logging Statement Definition and Usage
The logging statement configures a
wide
variety of logging options for the name server. Its channel phrase
associates output methods, format options, and severity levels with
a name that can then be used with the category phrase
to select how various classes of messages are logged.
Only one logging statement is used to
define
as many channels and categories as desired. If there is no logging statement,
the logging configuration is:
logging {
category default { default_syslog; default_debug; };
category unmatched { null; };
};
If named is started with the
-L option, it logs to the specified file
at startup, instead of using syslog. In this case the logging
configuration is:
logging {
category default { default_logfile; default_debug; };
category unmatched { null; };
};
The logging configuration
is only established when
the entire configuration file has been parsed. When the server starts up, all logging messages
regarding syntax errors in the configuration file go to the default
channels, or to standard error if the -g option
was specified.
All log output goes to one or more channels;
there is no limit to the number of channels that can be created.
Every channel definition must include a destination clause that
says whether messages selected for the channel go to a file, go to a
particular syslog facility, go to the standard error stream, or are
discarded. The definition can optionally also limit the message severity level
that is accepted by the channel (the default is
info), and whether to include a
named-generated time stamp, the
category name,
and/or the severity level (the default is not to include any).
The null destination clause
causes all messages sent to the channel to be discarded;
in that case, other options for the channel are meaningless.
The file destination clause directs
the channel
to a disk file. It can include limitations
both on how large the file is allowed to become, and on how many
versions
of the file are saved each time the file is opened.
If the versions log file
option is used, then
named retains that many backup
versions of the file by
renaming them when opening. For example, to keep
three old versions
of the file lamers.log, just
before it is opened
lamers.log.1 is renamed to
lamers.log.2, lamers.log.0 is renamed
to lamers.log.1, and lamers.log is
renamed to lamers.log.0.
The versions unlimited option can be set to
not limit
the number of versions.
If a size option is associated with
the log file,
then renaming is only done when the file being opened exceeds the
indicated size. No backup versions are kept by default; any
existing
log file is simply appended.
The size option for files is used
to limit log
growth. If the file ever exceeds the size, then named
stops writing to the file unless it also has a versions option
associated with it. If backup versions are kept, the files are
rolled as
described above and a new one begun. If there is no
versions option, no more data is
written to the log
until some out-of-band mechanism removes or truncates the log to
less than the
maximum size. The default behavior is not to limit the size of
the
file.
Here is an example using the size and
versions options:
channel an_example_channel {
file "example.log" versions 3 size 20m;
print-time yes;
print-category yes;
};
The syslog destination clause
directs the
channel to the system log. Its argument is a
syslog facility as described in the syslog man
page. Known facilities are kern, user,
mail, daemon, auth,
syslog, lpr, news,
uucp, cron, authpriv,
ftp, local0, local1,
local2, local3, local4,
local5, local6, and
local7; however, not all facilities
are supported on
all operating systems.
How syslog handles messages
sent to
this facility is described in the syslog.conf man
page. On a system which uses a very old version of syslog, which
only uses two arguments to the openlog() function,
then this clause is silently ignored.
On Windows machines, syslog messages are directed to the EventViewer.
The severity clause works like syslog's
"priorities," except that they can also be used when writing
straight to a file rather than using syslog.
Messages which are not at least of the severity level given are
not selected for the channel; messages of higher severity
levels
are accepted.
When using syslog, the syslog.conf priorities
also determine what eventually passes through. For example,
defining a channel facility and severity as daemon and debug, but
only logging daemon.warning via syslog.conf,
causes messages of severity info and
notice to
be dropped. If the situation were reversed, with named writing
messages of only warning or higher,
then syslogd would
print all messages it received from the channel.
The stderr destination clause
directs the
channel to the server's standard error stream. This is intended
for
use when the server is running as a foreground process, as
when debugging a configuration, for example.
The server can supply extensive debugging information when
it is in debugging mode. If the server's global debug level is
greater
than zero, debugging mode is active. The global debug
level is set either by starting the named server
with the -d flag followed by a positive integer,
or by running rndc trace.
The global debug level
can be set to zero, and debugging mode turned off, by running rndc
notrace. All debugging messages in the server have a debug
level; higher debug levels give more detailed output. Channels
that specify a specific debug severity, for example:
channel specific_debug_level {
file "foo";
severity debug 3;
};
get debugging output of level 3 or less any time the
server is in debugging mode, regardless of the global debugging
level. Channels with dynamic
severity use the
server's global debug level to determine what messages to print.
If print-time is set to
yes, then the date and time are logged.
print-time may be specified for a
syslog channel, but is usually
unnecessary since syslog also logs
the date and time. If print-category is
set to yes, then the
category of the message is logged as well. Finally, if
print-severity is set, then the severity
level of the message is logged.
The print- options may
be used in any combination, and are always printed in the
following
order: time, category, severity. Here is an example where all
three print- options
are on:
28-Feb-2000 15:05:32.863 general: notice: running
If buffered has been turned on, the output
to files is not flushed after each log entry. By default
all log messages are flushed.
There are four predefined channels that are used for
named's default logging, as follows.
If named is started with the
-L, then a
fifth channel, default_logfile, is added.
How they are
used is described in the section called “The category Phrase”.
channel default_syslog {
// send to syslog's daemon facility
syslog daemon;
// only send priority info and higher
severity info;
};
channel default_debug {
// write to named.run in the working directory
// Note: stderr is used instead of "named.run" if
// the server is started with the '-g' option.
file "named.run";
// log at the server's current debug level
severity dynamic;
};
channel default_stderr {
// writes to stderr
stderr;
// only send priority info and higher
severity info;
};
channel null {
// toss anything sent to this channel
null;
};
channel default_logfile {
// this channel is only present if named is
// started with the -L option, whose argument
// provides the file name
file "...";
// log at the server's current debug level
severity dynamic;
};
The default_debug channel has the
special
property that it only produces output when the server's debug
level is
non-zero. It normally writes to a file called named.run
in the server's working directory.
For security reasons, when the -u
command-line option is used, the named.run file
is created only after named has
changed to the
new UID, and any debug output generated while named is
starting - and still running as root - is discarded.
To capture this output, run the server with the -L
option to specify a default logfile, or the -g
option to log to standard error which can be redirected to a file.
Once a channel is defined, it cannot be redefined.
The built-in channels cannot be altered directly, but
the default logging can be modified by pointing categories at defined channels.
There are many categories, so desired logs
can be sent anywhere while unwanted logs are ignored. If
a list of channels is not specified for a category, log
messages
in that category are sent to the default category
instead. If no default category is specified, the following
"default default" is used:
category default { default_syslog; default_debug; };
If named is started with the
-L option, the default category is:
category default { default_logfile; default_debug; };
As an example, let's say a user wants to log security events to
a file, but also wants to keep the default logging behavior. They would
specify the following:
channel my_security_channel {
file "my_security_file";
severity info;
};
category security {
my_security_channel;
default_syslog;
default_debug;
};
To discard all messages in a category, specify the null channel:
category xfer-out { null; };
category notify { null; };
The following are the available categories and brief descriptions
of the types of log information they contain. More
categories may be added in future BIND releases.
The query-errors Category
The query-errors category is
used to indicate why and how specific queries resulted in
responses which indicate an error. Normally, these messages
will be logged at debug logging levels;
note, however, that if query logging is active, some are
logged at info. The logging levels are
described below:
At debug level 1 or higher - or at
info, when query logging is active - each
response with response code SERVFAIL is logged as follows:
client 127.0.0.1#61502: query failed (SERVFAIL) for www.example.com/IN/AAAA at query.c:3880
This means an error resulting in SERVFAIL was detected at line
3880 of source file query.c. Log messages
of this level are particularly helpful in identifying the cause of
SERVFAIL for an authoritative server.
At debug level 2 or higher, detailed
context information about recursive resolutions that resulted in
SERVFAIL is logged. The log message looks like this:
fetch completed at resolver.c:2970 for www.example.com/A
in 10.000183: timed out/success [domain:example.com,
referral:2,restart:7,qrysent:8,timeout:5,lame:0,quota:0,neterr:0,
badresp:1,adberr:0,findfail:0,valfail:0]
The first part before the colon shows that a recursive
resolution for AAAA records of www.example.com completed
in 10.000183 seconds, and the final result that led to the
SERVFAIL was determined at line 2970 of source file
resolver.c.
The next part shows the detected final result and the
latest result of DNSSEC validation. The latter is always
"success" when no validation attempt was made. In this example,
this query probably resulted in SERVFAIL because all name
servers are down or unreachable, leading to a timeout in 10
seconds. DNSSEC validation was probably not attempted.
The last part, enclosed in square brackets, shows statistics
collected for this particular resolution attempt.
The domain field shows the deepest zone that
the resolver reached; it is the zone where the error was
finally detected. The meaning of the other fields is
summarized in the following table.
At debug level 3 or higher, the same
messages as those at debug level 1 are
logged for errors other than SERVFAIL. Note that negative
responses such as NXDOMAIN are not errors, and are not logged
at this debug level.
At debug level 4 or higher, the
detailed context information logged at debug
level 2 is logged for errors other than SERVFAIL and
for negative responses such as NXDOMAIN.
options Statement Definition and
Usage
The options statement sets up global
options
to be used by BIND. This statement
may appear only
once in a configuration file. If there is no options
statement, an options block with each option set to its default is
used.
- attach-cache
-
This option allows multiple views to share a single cache
database.
Each view has its own cache database by default, but
if multiple views have the same operational policy
for name resolution and caching, those views can
share a single cache to save memory, and possibly
improve resolution efficiency, by using this option.
The attach-cache option
may also be specified in view
statements, in which case it overrides the
global attach-cache option.
The cache_name specifies
the cache to be shared.
When the named server configures
views which are supposed to share a cache, it
creates a cache with the specified name for the
first view of these sharing views.
The rest of the views simply refer to the
already-created cache.
One common configuration to share a cache is to
allow all views to share a single cache.
This can be done by specifying
attach-cache as a global
option with an arbitrary name.
Another possible operation is to allow a subset of
all views to share a cache while the others
retain their own caches.
For example, if there are three views A, B, and C,
and only A and B should share a cache, specify the
attach-cache option as a view of A (or
B)'s option, referring to the other view name:
view "A" {
// this view has its own cache
...
};
view "B" {
// this view refers to A's cache
attach-cache "A";
};
view "C" {
// this view has its own cache
...
};
Views that share a cache must have the same policy
on configurable parameters that may affect caching.
The current implementation requires the following
configurable options be consistent among these
views:
check-names,
cleaning-interval,
dnssec-accept-expired,
dnssec-validation,
max-cache-ttl,
max-ncache-ttl,
max-cache-size, and
zero-no-soa-ttl.
Note that there may be other parameters that may
cause confusion if they are inconsistent for
different views that share a single cache.
For example, if these views define different sets of
forwarders that can return different answers for the
same question, sharing the answer does not make
sense or could even be harmful.
It is administrator's responsibility to ensure that
configuration differences in different views do
not cause disruption with a shared cache.
- directory
This sets the working directory of the server.
Any non-absolute pathnames in the configuration file are
taken as relative to this directory. The default
location for most server output files
(e.g., named.run) is this directory.
If a directory is not specified, the working directory
defaults to ".", the directory from
which the server was started. The directory specified
should be an absolute path. It is
strongly recommended
that the directory be writable by the effective user
ID of the named process.
- dnstap
-
dnstap is a fast, flexible method
for capturing and logging DNS traffic. Developed by
Robert Edmonds at Farsight Security, Inc., and supported
by multiple DNS implementations, dnstap
uses
libfstrm (a lightweight high-speed
framing library, see
https://github.com/farsightsec/fstrm) to send
event payloads which are encoded using Protocol Buffers
(libprotobuf-c, a mechanism for
serializing structured data developed
by Google, Inc.; see
https://developers.google.com/protocol-buffers).
To enable dnstap at compile time,
the fstrm and protobuf-c
libraries must be available, and BIND must be configured with
--enable-dnstap.
The dnstap option is a bracketed list
of message types to be logged. These may be set differently
for each view. Supported types are client,
auth, resolver, and
forwarder. Specifying type
all causes all dnstap
messages to be logged, regardless of type.
Each type may take an additional argument to indicate whether
to log query messages or
response messages; if not specified,
both queries and responses are logged.
Example: To log all authoritative queries and responses,
recursive client responses, and upstream queries sent by
the resolver, use:
dnstap {
auth;
client response;
resolver query;
};
Logged dnstap messages can be parsed
using the dnstap-read utility (see
dnstap-read(1) for details).
For more information on dnstap, see
http://dnstap.info.
The fstrm library has a number of tunables that are exposed
in named.conf, and can be modified
if necessary to improve performance or prevent loss of data.
These are:
-
fstrm-set-buffer-hint: The
threshold number of bytes to accumulate in the output
buffer before forcing a buffer flush. The minimum is
1024, the maximum is 65536, and the default is 8192.
-
fstrm-set-flush-timeout: The number
of seconds to allow unflushed data to remain in the
output buffer. The minimum is 1 second, the maximum is
600 seconds (10 minutes), and the default is 1 second.
-
fstrm-set-output-notify-threshold:
The number of outstanding queue entries to allow on
an input queue before waking the I/O thread.
The minimum is 1 and the default is 32.
-
fstrm-set-output-queue-model:
The queuing semantics to use for queue
objects. The default is
mpsc
(multiple producer, single consumer); the other
option is spsc (single producer,
single consumer).
-
fstrm-set-input-queue-size: The
number of queue entries to allocate for each
input queue. This value must be a power of 2.
The minimum is 2, the maximum is 16384, and
the default is 512.
-
fstrm-set-output-queue-size:
The number of queue entries to allocate for each
output queue. The minimum is 2, the maximum is
system-dependent and based on
IOV_MAX,
and the default is 64.
-
fstrm-set-reopen-interval:
The number of seconds to wait between attempts to
reopen a closed output stream. The minimum is 1 second,
the maximum is 600 seconds (10 minutes), and the default
is 5 seconds.
Note that all of the above minimum, maximum, and default
values are set by the libfstrm library,
and may be subject to change in future versions of the
library. See the libfstrm documentation
for more information.
- dnstap-output
-
This configures the path to which the dnstap
frame stream is sent if dnstap
is enabled at compile time and active.
The first argument is either file or
unix, indicating whether the destination
is a file or a Unix domain socket. The second argument
is the path of the file or socket. (Note: when using a
socket, dnstap messages are
only sent if another process such as
fstrm_capture
(provided with libfstrm) is listening on
the socket.)
dnstap-output can only be set globally
in options. Currently, it can only be
set once while named is running;
once set, it cannot be changed by
rndc reload or
rndc reconfig.
- dnstap-identity
This specifies an identity string to send in
dnstap messages. If set to
hostname, which is the default, the
server's hostname is sent. If set to
none, no identity string is sent.
- dnstap-version
This specifies a version string to send in
dnstap messages. The default is the
version number of the BIND release. If set to
none, no version string is sent.
- geoip-directory
When named is compiled using the
MaxMind GeoIP2 geolocation API, or the legacy GeoIP API,
this specifies the directory containing GeoIP
database files. By default, the option is set based on
the prefix used to build the libmaxminddb
module; for example, if the library is installed in
/usr/local/lib, then the default
geoip-directory is
/usr/local/share/GeoIP. On Windows,
the default is the named working
directory. See the section called “acl Statement Definition and
Usage” for details about
geoip ACLs.
- key-directory
This is the
directory where the public and private DNSSEC key files
should be found when performing a dynamic update of secure zones, if different than the current working
directory. (Note that this option has no effect on the
paths for files containing non-DNSSEC keys such as
bind.keys,
rndc.key, or
session.key.)
- lmdb-mapsize
-
When named is built with liblmdb,
this option sets a maximum size for the memory map of
the new-zone database (NZD) in LMDB database format.
This database is used to store configuration information
for zones added using rndc addzone.
Note that this is not the NZD database file size, but
the largest size that the database may grow to.
Because the database file is memory mapped, its size is
limited by the address space of the named process. The
default of 32 megabytes was chosen to be usable with
32-bit named builds. The largest
permitted value is 1 terabyte. Given typical zone
configurations without elaborate ACLs, a 32 MB NZD file
ought to be able to hold configurations of about 100,000
zones.
- managed-keys-directory
-
This specifies the directory in which to store the files that
track managed DNSSEC keys. By default, this is the working
directory. The directory must
be writable by the effective user ID of the
named process.
If named is not configured to use views,
managed keys for the server are tracked in a single
file called managed-keys.bind.
Otherwise, managed keys are tracked in separate files,
one file per view; each file name is the view name
(or, if it contains characters that are incompatible with
use as a file name, the SHA256 hash of the view name),
followed by the extension
.mkeys.
(Note: in earlier releases, file names for views
always used the SHA256 hash of the view name. To ensure
compatibility after upgrading, if a file using the old
name format is found to exist, it is used instead
of the new format.)
- named-xfer
This option is obsolete.
In BIND 9, no separate
named-xfer program is needed;
its functionality is built into the name server.
- tkey-gssapi-keytab
This is the KRB5 keytab file to use for GSS-TSIG updates. If
this option is set and tkey-gssapi-credential is not
set, updates are allowed with any key
matching a principal in the specified keytab.
- tkey-gssapi-credential
This is the security credential with which the server should
authenticate keys requested by the GSS-TSIG protocol.
Currently only Kerberos 5 authentication is available;
the credential is a Kerberos principal which the
server can acquire through the default system key
file, normally /etc/krb5.keytab.
The location of the keytab file can be overridden using the
tkey-gssapi-keytab option. Normally this principal is
of the form "DNS/server.domain".
To use GSS-TSIG, tkey-domain must
also be set if a specific keytab is not set with
tkey-gssapi-keytab.
- tkey-domain
This domain is appended to the names of all shared keys
generated with TKEY. When a
client requests a TKEY exchange,
it may or may not specify the desired name for the
key. If present, the name of the shared key is
client-specified part +
tkey-domain. Otherwise, the
name of the shared key is random hex
digits + tkey-domain.
In most cases, the domainname
should be the server's domain name, or an otherwise
nonexistent subdomain like
"_tkey.domainname". If
using GSS-TSIG, this variable must be defined, unless
a specific keytab is specified using tkey-gssapi-keytab.
- tkey-dhkey
This is the Diffie-Hellman key used by the server
to generate shared keys with clients using the Diffie-Hellman
mode
of TKEY. The server must be
able to load the
public and private keys from files in the working directory.
In
most cases, the key_name should be the server's host name.
- cache-file
This is for testing only. Do not use.
- dump-file
This is the pathname of the file the server dumps
the database to, when instructed to do so with
rndc dumpdb.
If not specified, the default is named_dump.db.
- memstatistics-file
This is the pathname of the file the server writes memory
usage statistics to on exit. If not specified,
the default is named.memstats.
- lock-file
-
This is the pathname of a file on which named
attempts to acquire a file lock when starting for
the first time; if unsuccessful, the server
terminates, under the assumption that another
server is already running. If not specified, the default is
none.
Specifying lock-file none disables the
use of a lock file. lock-file is
ignored if named was run using the -X
option, which overrides it. Changes to
lock-file are ignored if
named is being reloaded or
reconfigured; it is only effective when the server is
first started.
- pid-file
This is the pathname of the file the server writes its process ID
in. If not specified, the default is
/var/run/named/named.pid.
The PID file is used by programs that send signals to
the running
name server. Specifying pid-file none disables the
use of a PID file; no file is written and any
existing one is removed. Note that none
is a keyword, not a filename, and therefore is not enclosed
in
double quotes.
- recursing-file
This is the pathname of the file where the server dumps
the queries that are currently recursing, when instructed
to do so with rndc recursing.
If not specified, the default is named.recursing.
- statistics-file
This is the pathname of the file the server appends statistics
to, when instructed to do so using rndc stats.
If not specified, the default is named.stats in the
server's current directory. The format of the file is
described
in the section called “The Statistics File”.
- bindkeys-file
This is the pathname of a file to override the built-in trusted
keys provided by named.
See the discussion of dnssec-validation
for details. If not specified, the default is
/etc/bind.keys.
- secroots-file
This is the pathname of the file the server dumps
security roots to, when instructed to do so with
rndc secroots.
If not specified, the default is
named.secroots.
- session-keyfile
This is the pathname of the file into which to write a TSIG
session key generated by named for use by
nsupdate -l. If not specified, the
default is /var/run/named/session.key.
(See the section called “Dynamic Update Policies”, and in
particular the discussion of the
update-policy statement's
local option for more
information about this feature.)
- session-keyname
This is the key name to use for the TSIG session key.
If not specified, the default is local-ddns.
- session-keyalg
This is the algorithm to use for the TSIG session key.
Valid values are hmac-sha1, hmac-sha224, hmac-sha256,
hmac-sha384, hmac-sha512, and hmac-md5. If not
specified, the default is hmac-sha256.
- port
This is the UDP/TCP port number the server uses to
receive and send DNS protocol traffic.
The default is 53. This option is mainly intended for server
testing;
a server using a port other than 53 is not able to
communicate with
the global DNS.
- dscp
This is the global Differentiated Services Code Point (DSCP)
value to classify outgoing DNS traffic, on operating
systems that support DSCP. Valid values are 0 through 63.
It is not configured by default.
- random-device
This specifies a source of entropy to be used by the server. Entropy is
primarily needed
for DNSSEC operations, such as TKEY transactions and dynamic
update of signed
zones. This option specifies the device (or file) from which
to read
entropy. If it is a file, operations requiring entropy will
fail when the
file has been exhausted. If random-device is not specified, the default value
is
/dev/random
(or equivalent) when present, and none otherwise. The
random-device option takes
effect during
the initial configuration load at server startup time and
is ignored on subsequent reloads.
- preferred-glue
If specified, the listed type (A or AAAA) is emitted
before other glue
in the additional section of a query response.
The default is to prefer A records when responding
to queries that arrived via IPv4 and AAAA when
responding to queries that arrived via IPv6.
-
root-delegation-only
-
This turns on enforcement of delegation-only in TLDs
(top-level domains) and root zones with an optional
exclude list.
DS queries are expected to be made to and be answered by
delegation-only zones. Such queries and responses are
treated as an exception to delegation-only processing
and are not converted to NXDOMAIN responses, provided
a CNAME is not discovered at the query name.
If a delegation-only zone server also serves a child
zone, it is not always possible to determine whether
an answer comes from the delegation-only zone or the
child zone. SOA NS and DNSKEY records are apex-only
records and a matching response that contains
these records or DS is treated as coming from a
child zone. RRSIG records are also examined to see
if they are signed by a child zone, and the
authority section is examined to see if there
is evidence that the answer is from the child zone.
Answers that are determined to be from a child zone
are not converted to NXDOMAIN responses. Despite
all these checks, there is still a possibility of
false negatives when a child zone is being served.
Similarly, false positives can arise from empty nodes
(no records at the name) in the delegation-only zone
when the query type is not ANY.
Note that some TLDs are not delegation-only; e.g., "DE", "LV",
"US", and "MUSEUM". This list is not exhaustive.
options {
root-delegation-only exclude { "de"; "lv"; "us"; "museum"; };
};
- disable-algorithms
-
This disables the specified DNSSEC algorithms at and below the
specified name.
Multiple disable-algorithms
statements are allowed.
Only the best-match disable-algorithms
clause is used to determine the algorithms.
If all supported algorithms are disabled, the zones covered
by the disable-algorithms setting are treated
as insecure.
Configured trust anchors in trusted-keys
or managed-keys that match a disabled
algorithm are ignored and treated as if they were not
configured.
- disable-ds-digests
-
This disables the specified DS digest types at and below the
specified name.
Multiple disable-ds-digests
statements are allowed.
Only the best-match disable-ds-digests
clause is used to determine the digest types.
If all supported digest types are disabled, the zones covered
by disable-ds-digests are treated
as insecure.
- dnssec-lookaside
-
When set, dnssec-lookaside provides the
validator with an alternate method to validate DNSKEY
records at the top of a zone. When a DNSKEY is at or
below a domain specified by the deepest
dnssec-lookaside, and the normal DNSSEC
validation has left the key untrusted, the trust-anchor
is appended to the key name and a DLV record is
looked up to see if it can validate the key. If the DLV
record validates a DNSKEY (similarly to the way a DS
record does), the DNSKEY RRset is deemed to be trusted.
If dnssec-lookaside is set to
no, then dnssec-lookaside
is not used.
Note: the ISC-provided DLV service at
dlv.isc.org has been shut down.
The dnssec-lookaside auto;
configuration option, which set named
to use ISC DLV with minimal configuration, has
accordingly been removed.
- dnssec-must-be-secure
This specifies hierarchies which must be or may not be secure
(signed and validated). If yes,
then named only accepts answers if
they are secure. If no, then normal
DNSSEC validation applies, allowing insecure answers to
be accepted. The specified domain must be under a
trusted-keys or
managed-keys statement, or
dnssec-validation auto must be active.
- dns64
-
This directive instructs named to
return mapped IPv4 addresses to AAAA queries when
there are no AAAA records. It is intended to be
used in conjunction with a NAT64. Each
dns64 defines one DNS64 prefix.
Multiple DNS64 prefixes can be defined.
Compatible IPv6 prefixes have lengths of 32, 40, 48, 56,
64, and 96, per RFC 6052. Bits 64..71 inclusive must
be zero, with the most significant bit of the prefix in
position 0.
In addition, a reverse IP6.ARPA zone is created for
the prefix to provide a mapping from the IP6.ARPA names
to the corresponding IN-ADDR.ARPA names using synthesized
CNAMEs. dns64-server and
dns64-contact can be used to specify
the name of the server and contact for the zones. These
can be set at the view/options level but not
on a per-prefix basis.
Each dns64 supports an optional
clients ACL that determines which
clients are affected by this directive. If not defined,
it defaults to any;.
Each dns64 supports an optional
mapped ACL that selects which
IPv4 addresses are to be mapped in the corresponding
A RRset. If not defined, it defaults to
any;.
Normally, DNS64 does not apply to a domain name that
owns one or more AAAA records; these records are
simply returned. The optional
exclude ACL allows specification
of a list of IPv6 addresses that are ignored
if they appear in a domain name's AAAA records;
DNS64 is applied to any A records the domain
name owns. If not defined, exclude
defaults to ::ffff:0.0.0.0/96.
A optional suffix can also
be defined to set the bits trailing the mapped
IPv4 address bits. By default these bits are
set to ::. The bits
matching the prefix and mapped IPv4 address
must be zero.
If recursive-only is set to
yes, the DNS64 synthesis
only happens for recursive queries. The default
is no.
If break-dnssec is set to
yes, the DNS64 synthesis
happens even if the result, if validated, would
cause a DNSSEC validation failure. If this option
is set to no (the default), the DO
is set on the incoming query, and there are RRSIGs on
the applicable records, then synthesis does not happen.
acl rfc1918 { 10/8; 192.168/16; 172.16/12; };
dns64 64:FF9B::/96 {
clients { any; };
mapped { !rfc1918; any; };
exclude { 64:FF9B::/96; ::ffff:0000:0000/96; };
suffix ::;
};
- dnssec-loadkeys-interval
When a zone is configured with auto-dnssec
maintain;, its key repository must be checked
periodically to see if any new keys have been added
or any existing keys' timing metadata has been updated
(see dnssec-keygen(8) and
dnssec-settime(8)). The
dnssec-loadkeys-interval option
sets the frequency of automatic repository checks, in
minutes. The default is 60 (1 hour),
the minimum is 1 (1 minute), and the
maximum is 1440 (24 hours); any higher
value is silently reduced.
- dnssec-update-mode
-
If this option is set to its default value of
maintain in a zone of type
master which is DNSSEC-signed
and configured to allow dynamic updates (see
the section called “Dynamic Update Policies”), and
if named has access to the
private signing key(s) for the zone, then
named automatically signs all new
or changed records and maintains signatures for the zone
by regenerating RRSIG records whenever they approach
their expiration date.
If the option is changed to no-resign,
then named signs all new or
changed records, but scheduled maintenance of
signatures is disabled.
With either of these settings, named
rejects updates to a DNSSEC-signed zone when the
signing keys are inactive or unavailable to
named. (A planned third option,
external, will disable all automatic
signing and allow DNSSEC data to be submitted into a zone
via dynamic update; this is not yet implemented.)
- nta-lifetime
-
This specifies the default lifetime, in seconds,
for negative trust anchors added
via rndc nta.
A negative trust anchor selectively disables
DNSSEC validation for zones that are known to be
failing because of misconfiguration, rather than
an attack. When data to be validated is
at or below an active NTA (and above any other
configured trust anchors), named
aborts the DNSSEC validation process and treats the data as
insecure rather than bogus. This continues until the
NTA's lifetime is elapsed. NTAs persist
across named restarts.
For convenience, TTL-style time-unit suffixes can be
used to specify the NTA lifetime in seconds, minutes,
or hours. nta-lifetime defaults to
one hour; it cannot exceed one week.
- nta-recheck
-
This specifies how often to check whether negative
trust anchors added via rndc nta
are still necessary.
A negative trust anchor is normally used when a
domain has stopped validating due to operator error;
it temporarily disables DNSSEC validation for that
domain. In the interest of ensuring that DNSSEC
validation is turned back on as soon as possible,
named periodically sends a
query to the domain, ignoring negative trust anchors,
to find out whether it can now be validated. If so,
the negative trust anchor is allowed to expire early.
Validity checks can be disabled for an individual
NTA by using rndc nta -f, or
for all NTAs by setting nta-recheck
to zero.
For convenience, TTL-style time-unit suffixes can be
used to specify the NTA recheck interval in seconds,
minutes, or hours. The default is five minutes. It
cannot be longer than nta-lifetime,
which cannot be longer than a week.
- max-zone-ttl
-
This specifies a maximum permissible TTL value in seconds.
For convenience, TTL-style time-unit suffixes may be
used to specify the maximum value.
When loading a zone file using a
masterfile-format of
text or raw,
any record encountered with a TTL higher than
max-zone-ttl causes the zone to
be rejected.
This is useful in DNSSEC-signed zones because when
rolling to a new DNSKEY, the old key needs to remain
available until RRSIG records have expired from
caches. The max-zone-ttl option guarantees
that the largest TTL in the zone is no higher
than the set value.
(Note: because map-format files
load directly into memory, this option cannot be
used with them.)
The default value is unlimited.
A max-zone-ttl of zero is treated as
unlimited.
- serial-update-method
-
Zones configured for dynamic DNS may use this
option to set the update method to be used for
the zone serial number in the SOA record.
With the default setting of
serial-update-method increment;, the
SOA serial number is incremented by one each time
the zone is updated.
When set to
serial-update-method unixtime;, the
SOA serial number is set to the number of seconds
since the Unix epoch, unless the serial number is
already greater than or equal to that value, in which
case it is simply incremented by one.
When set to
serial-update-method date;, the
new SOA serial number is the current date
in the form "YYYYMMDD", followed by two zeroes,
unless the existing serial number is already greater
than or equal to that value, in which case it is
incremented by one.
- zone-statistics
-
If full, the server collects
statistical data on all zones, unless specifically
turned off on a per-zone basis by specifying
zone-statistics terse or
zone-statistics none
in the zone statement.
The default is terse, providing
minimal statistics on zones (including name and
current serial number, but not query type
counters).
These statistics may be accessed via the
statistics-channel or
using rndc stats, which
dumps them to the file listed
in the statistics-file. See
also the section called “The Statistics File”.
For backward compatibility with earlier versions
of BIND 9, the zone-statistics
option can also accept yes
or no; yes
has the same meaning as full.
As of BIND 9.10,
no has the same meaning
as none; previously, it
was the same as terse.
- automatic-interface-scan
-
If yes and supported by the operating
system, this automatically rescans network interfaces when the
interface addresses are added or removed. The default is
yes. This configuration option does
not affect the time-based interface-interval
option; it is recommended to set the time-based
interface-interval to 0 when the operator
confirms that automatic interface scanning is supported by the
operating system.
The automatic-interface-scan implementation
uses routing sockets for the network interface discovery;
therefore, the operating system must support the routing
sockets for this feature to work.
- allow-new-zones
-
If yes, then zones can be
added at runtime via rndc addzone.
The default is no.
Newly added zones' configuration parameters
are stored so that they can persist after the
server is restarted. The configuration information
is saved in a file called
viewname.nzf
(or, if named is compiled with
liblmdb, in an LMDB database file called
viewname.nzd).
viewname is the name of the
view, unless the view name contains characters that are
incompatible with use as a file name, in which case a
cryptographic hash of the view name is used instead.
Configurations for zones added at runtime are
stored either in a new-zone file (NZF) or a new-zone
database (NZD), depending on whether
named was linked with
liblmdb at compile time.
See rndc(8) for further details
about rndc addzone.
- auth-nxdomain
If yes, then the AA bit
is always set on NXDOMAIN responses, even if the server is
not actually
authoritative. The default is no.
- deallocate-on-exit
This option was used in BIND
8 to enable checking
for memory leaks on exit. BIND 9 ignores the option and always performs
the checks.
- memstatistics
This writes memory statistics to the file specified by
memstatistics-file at exit.
The default is no unless
-m record is specified on the command line, in
which case it is yes.
- dialup
-
If yes, then the
server treats all zones as if they are doing zone transfers
across
a dial-on-demand dialup link, which can be brought up by
traffic
originating from this server. Although this setting has different effects
according
to zone type, it concentrates the zone maintenance so that
everything
happens quickly, once every heartbeat-interval,
ideally during a single call. It also suppresses some
normal
zone maintenance traffic. The default is no.
If specified in the view and
zone statements, the dialup option
overrides the global dialup option.
If the zone is a primary zone, the server sends out a
NOTIFY
request to all the secondaries (default). This should trigger the
zone serial
number check in the secondary (providing it supports NOTIFY),
allowing the secondary
to verify the zone while the connection is active.
The set of servers to which NOTIFY is sent can be controlled
by
notify and also-notify.
If the
zone is a secondary or stub zone, the server suppresses
the regular
"zone up to date" (refresh) queries and only performs them
when the
heartbeat-interval expires, in
addition to sending
NOTIFY requests.
Finer control can be achieved by using
notify, which only sends NOTIFY
messages;
notify-passive, which sends NOTIFY
messages and
suppresses the normal refresh queries; refresh,
which suppresses normal refresh processing and sends refresh
queries
when the heartbeat-interval
expires; and
passive, which disables normal
refresh
processing.
Note that normal NOTIFY processing is not affected by
dialup.
- fake-iquery
In BIND 8, this option
enabled simulating the obsolete DNS query type
IQUERY. BIND 9 never does
IQUERY simulation.
- fetch-glue
This option is obsolete.
In BIND 8, fetch-glue yes
caused the server to attempt to fetch glue resource records
it
did not have when constructing the additional
data section of a response. This is now considered a bad
idea
and BIND 9 never does it.
- flush-zones-on-shutdown
When the nameserver exits upon receiving SIGTERM,
flush or do not flush any pending zone writes. The default
is
flush-zones-on-shutdown no.
- geoip-use-ecs
When BIND is compiled with GeoIP support and configured
with "geoip" ACL elements, this option indicates whether
the EDNS Client Subnet option, if present in a request,
should be used for matching against the GeoIP database.
The default is
geoip-use-ecs yes.
- has-old-clients
This option was incorrectly implemented
in BIND 8, and is ignored by BIND 9.
To achieve the intended effect
of
has-old-clients yes, specify
the two separate options auth-nxdomain yes
and rfc2308-type1 no instead.
- host-statistics
In BIND 8, this enabled keeping of
statistics for every host that the name server interacts
with.
It is not implemented in BIND 9.
- root-key-sentinel
If yes, respond to root key sentinel probes as described in
draft-ietf-dnsop-kskroll-sentinel-08. The default is
yes.
- maintain-ixfr-base
This option is obsolete.
It was used in BIND 8 to
determine whether a transaction log was
kept for Incremental Zone Transfer. BIND 9 maintains a transaction
log whenever possible. To disable outgoing
incremental zone
transfers, use provide-ixfr no.
- message-compression
If yes, DNS name compression is
used in responses to regular queries (not including
AXFR or IXFR, which always use compression). Setting
this option to no reduces CPU
usage on servers and may improve throughput. However,
it increases response size, which may cause more queries
to be processed using TCP; a server with compression
disabled is out of compliance with RFC 1123 Section
6.1.3.2. The default is yes.
- minimal-responses
-
If set to yes, then when generating
responses the server only adds records to the authority
and additional data sections when they are required (e.g.
delegations, negative responses). This may improve the
performance of the server.
When set to no-auth, the
server omits records from the authority section
unless they are required, but it may still add
records to the additional section. When set to
no-auth-recursive, this
is only done if the query is recursive. These
settings are useful when answering stub clients,
which usually ignore the authority section.
no-auth-recursive is
designed for mixed-mode servers that handle
both authoritative and recursive queries.
The default is no.
- minimal-any
If set to yes, the server replies with only one
of the RRsets for the query name, and its covering
RRSIGs if any, when
generating a positive response to a query of type
ANY over UDP, instead of replying with all known
RRsets for the name. Similarly, a query for type
RRSIG is answered with the RRSIG records covering
only one type. This can reduce the impact of some kinds
of attack traffic, without harming legitimate
clients. (Note, however, that the RRset returned is the
first one found in the database; it is not necessarily
the smallest available RRset.)
Additionally, minimal-responses is
turned on for these queries, so no unnecessary records
are added to the authority or additional sections.
The default is no.
- multiple-cnames
This option was used in BIND 8 to allow
a domain name to have multiple CNAME records, in violation of
the DNS standards. BIND 9.2 onwards
always strictly enforces the CNAME rules both in primary
files and dynamic updates.
- notify
-
If yes (the default),
DNS NOTIFY messages are sent when a zone the server is
authoritative for
changes; see the section called “Notify”. The messages are
sent to the
servers listed in the zone's NS records (except the primary
server identified
in the SOA MNAME field), and to any servers listed in the
also-notify option.
If master-only, notifies are only
sent
for primary zones.
If explicit, notifies are sent only
to
servers explicitly listed using also-notify.
If no, no notifies are sent.
The notify option may also be
specified in the zone
statement,
in which case it overrides the options notify statement.
It would only be necessary to turn off this option if it
caused secondary zones
to crash.
- notify-to-soa
If yes, do not check the name servers
in the NS RRset against the SOA MNAME. Normally a NOTIFY
message is not sent to the SOA MNAME (SOA ORIGIN), as it is
supposed to contain the name of the ultimate primary server.
Sometimes, however, a secondary server is listed as the SOA MNAME in
hidden primary configurations; in that case,
the ultimate primary should be set to still send NOTIFY messages to
all the name servers listed in the NS RRset.
- recursion
If yes, and a
DNS query requests recursion, then the server attempts
to do
all the work required to answer the query. If recursion is
off
and the server does not already know the answer, it
returns a
referral response. The default is
yes.
Note that setting recursion no does not prevent
clients from getting data from the server's cache; it only
prevents new data from being cached as an effect of client
queries.
Caching may still occur as an effect the server's internal
operation, such as NOTIFY address lookups.
- request-nsid
If yes, then an empty EDNS(0)
NSID (Name Server Identifier) option is sent with all
queries to authoritative name servers during iterative
resolution. If the authoritative server returns an NSID
option in its response, then its contents are logged in
the resolver category at level
info.
The default is no.
- request-sit
This experimental option is obsolete.
- require-server-cookie
-
If yes, require a valid server cookie before sending a full
response to a UDP request from a cookie-aware client.
BADCOOKIE is sent if there is a bad or nonexistent
server cookie.
The default is no.
Users wishing to test that DNS COOKIE clients correctly handle BADCOOKIE, or who are
getting a lot of forged DNS requests with DNS COOKIES
present, should set this to yes.
Setting this to yes
results in a reduced amplification effect in a reflection
attack, as the BADCOOKIE response is smaller than
a full response, while also requiring a legitimate client
to follow up with a second query with the new, valid, cookie.
- answer-cookie
-
When set to the default value of yes,
COOKIE EDNS options are sent when applicable in
replies to client queries. If set to
no, COOKIE EDNS options are not
sent in replies. This can only be set at the global
options level, not per-view.
answer-cookie no is only intended as a
temporary measure, for use when named
shares an IP address with other servers that do not yet
support DNS COOKIE. A mismatch between servers on the
same address is not expected to cause operational
problems, but the option to disable COOKIE responses so
that all servers have the same behavior is provided out
of an abundance of caution. DNS COOKIE is an important
security mechanism, and should not be disabled unless
absolutely necessary.
- send-cookie
If yes, then a COOKIE EDNS
option is sent along with the query. If the
resolver has previously communicated with the server, the
COOKIE returned in the previous transaction is sent.
This is used by the server to determine whether
the resolver has talked to it before. A resolver
sending the correct COOKIE is assumed not to be an
off-path attacker sending a spoofed-source query;
the query is therefore unlikely to be part of a
reflection/amplification attack, so resolvers
sending a correct COOKIE option are not subject to
response rate limiting (RRL). Resolvers which
do not send a correct COOKIE option may be limited
to receiving smaller responses via the
nocookie-udp-size option.
The default is yes.
- nocookie-udp-size
This sets the maximum size of UDP responses that are
sent to queries without a valid server COOKIE. A value
below 128 is silently raised to 128. The default
value is 4096, but the max-udp-size
option may further limit the response size as the default
for max-udp-size is 1232.
- sit-secret
This experimental option is obsolete.
- cookie-algorithm
This sets the algorithm to be used when generating the
server cookie; the options are "aes", "sha1", or "sha256".
The default is "aes" if supported by the cryptographic
library; otherwise, "sha256".
- cookie-secret
-
If set, this is a shared secret used for generating
and verifying EDNS COOKIE options
within an anycast cluster. If not set, the system
generates a random secret at startup. The
shared secret is encoded as a hex string and needs
to be 128 bits for AES128, 160 bits for SHA1, and
256 bits for SHA256.
If there are multiple secrets specified, the first
one listed in named.conf is
used to generate new server cookies. The others
are only used to verify returned cookies.
- rfc2308-type1
-
Setting this to yes
causes the server to send NS records along with the SOA
record for negative
answers. The default is no.
Note
This is not yet implemented in BIND
9.
- trust-anchor-telemetry
-
This causes named to send specially formed
queries once per day to domains for which trust anchors
have been configured via trusted-keys,
managed-keys, or
dnssec-validation auto.
The query name used for these queries has the
form "_ta-xxxx(-xxxx)(...)".<domain>, where
each "xxxx" is a group of four hexadecimal digits
representing the key ID of a trusted DNSSEC key.
The key IDs for each domain are sorted smallest
to largest prior to encoding. The query type is NULL.
By monitoring these queries, zone operators are
able to see which resolvers have been updated to
trust a new key; this may help them decide when it
is safe to remove an old one.
The default is yes.
- use-id-pool
This option is obsolete.
BIND 9 always allocates query
IDs from a pool.
- use-ixfr
This option is obsolete.
To disable IXFR to a particular server or
servers, see
the information on the provide-ixfr option
in the section called “server Statement Definition and
Usage”.
See also
the section called “Incremental Zone Transfers (IXFR)”.
- provide-ixfr
See the description of
provide-ixfr in
the section called “server Statement Definition and
Usage”.
- request-ixfr
See the description of
request-ixfr in
the section called “server Statement Definition and
Usage”.
- request-expire
See the description of
request-expire in
the section called “server Statement Definition and
Usage”.
- treat-cr-as-space
This option was used in BIND
8 to make
the server treat carriage return ("\r") characters the same way
as a space or tab character,
to facilitate loading of zone files on a Unix system that
were generated
on an NT or DOS machine. In BIND 9, both UNIX "\n"
and NT/DOS "\r\n" newlines
are always accepted,
and the option is ignored.
-
additional-from-auth, additional-from-cache
-
These options control the behavior of an authoritative
server when
answering queries which have additional data, or when
following CNAME
and DNAME chains.
When both of these options are set to yes
(the default) and a
query is being answered from authoritative data (a zone
configured into the server), the additional data section of
the
reply is filled in using data from other authoritative
zones
and from the cache. In some situations this is undesirable,
such
as when there is concern over the correctness of the cache,
or
in servers where secondary zones may be added and modified by
untrusted third parties. Also, avoiding
the search for this additional data speeds up server
operations
at the possible expense of additional queries to resolve
what would
otherwise be provided in the additional section.
For example, if a query asks for an MX record for host foo.example.com,
and the record found is "MX 10 mail.example.net", normally the address
records (A and AAAA) for mail.example.net are provided as well,
if known, even though they are not in the example.com zone.
Setting these options to no
disables this behavior and makes
the server only search for additional data in the zone it
answers from.
These options are intended for use in authoritative-only
servers, or in authoritative-only views. Attempts to set
them to no without also
specifying
recursion no will cause the
server to
ignore the options and log a warning message.
Specifying additional-from-cache no actually
disables the use of the cache not only for additional data
lookups
but also when looking up the answer. This is usually the
desired
behavior in an authoritative-only server where the
correctness of
the cached data is an issue.
When a name server is non-recursively queried for a name
that is not
below the apex of any served zone, it normally answers with
an
"upwards referral" to the root servers or the servers of
some other
known parent of the query name. Since the data in an
upwards referral
comes from the cache, the server is not able to provide
upwards
referrals when additional-from-cache no
has been specified. Instead, it responds to such
queries
with REFUSED. This should not cause any problems since
upwards referrals are not required for the resolution
process.
- match-mapped-addresses
-
If yes, then an
IPv4-mapped IPv6 address matches any address-match
list entries that match the corresponding IPv4 address.
This option was introduced to work around a kernel quirk
in some operating systems that causes IPv4 TCP
connections, such as zone transfers, to be accepted on an
IPv6 socket using mapped addresses. This caused address-match
lists designed for IPv4 to fail to match. However,
named now solves this problem
internally. The use of this option is discouraged.
- filter-aaaa-on-v4
-
This option is only available when
BIND 9 is compiled with the
--enable-filter-aaaa option on the
"configure" command line. It is intended to help the
transition from IPv4 to IPv6 by not giving IPv6 addresses
to DNS clients unless they have connections to the IPv6
Internet. This is not recommended unless absolutely
necessary. The default is no.
The filter-aaaa-on-v4 option
may also be specified in view statements
to override the global filter-aaaa-on-v4
option.
If yes,
the DNS client is at an IPv4 address, in filter-aaaa,
and if the response does not include DNSSEC signatures,
then all AAAA records are deleted from the response.
This filtering applies to all responses and not only
authoritative responses.
If break-dnssec,
then AAAA records are deleted even when DNSSEC is enabled.
As suggested by the name, this causes the response to not verify,
because the DNSSEC protocol is designed to detect deletions.
This mechanism can erroneously cause other servers to
not give AAAA records to their clients.
A recursing server with both IPv6 and IPv4 network connections,
that queries an authoritative server using this mechanism
via IPv4, is denied AAAA records even if its client is
using IPv6.
This mechanism is applied to authoritative as well as
non-authoritative records.
A client using IPv4 that is not allowed recursion can
erroneously be given AAAA records because the server is not
allowed to check for A records.
Some AAAA records are given to IPv4 clients in glue records.
IPv4 clients that are servers can then erroneously
answer requests for AAAA records received via IPv4.
- filter-aaaa-on-v6
This is identical to filter-aaaa-on-v4,
except it filters AAAA responses to queries from IPv6
clients instead of IPv4 clients. To filter all
responses, set both options to yes.
- ixfr-from-differences
-
When yes and the server loads a new
version of a primary zone from its zone file or receives a
new version of a secondary file via zone transfer, it
compares the new version to the previous one and calculates
a set of differences. The differences are then logged in
the zone's journal file so that the changes can be
transmitted to downstream secondaries as an incremental zone
transfer.
By allowing incremental zone transfers to be used for
non-dynamic zones, this option saves bandwidth at the
expense of increased CPU and memory consumption at the
primary server.
In particular, if the new version of a zone is completely
different from the previous one, the set of differences
is of a size comparable to the combined size of the
old and new zone versions, and the server needs to
temporarily allocate memory to hold this complete
difference set.
ixfr-from-differences
also accepts master and
slave at the view and options
levels, which causes
ixfr-from-differences to be enabled for
all primary or secondary zones, respectively.
It is off by default.
Note: if inline signing is enabled for a zone, the
user-provided ixfr-from-differences
setting is ignored for that zone.
- multi-master
This should be set when there are multiple primary servers for a zone
and the
addresses refer to different machines. If yes, named does
not log
when the serial number on the primary is less than what named
currently
has. The default is no.
- auto-dnssec
-
Zones configured for dynamic DNS may use this
option to allow varying levels of automatic DNSSEC key
management. There are three possible settings:
auto-dnssec allow; permits
keys to be updated and the zone fully re-signed
whenever the user issues the command rndc sign
zonename.
auto-dnssec maintain; includes the
above, but also automatically adjusts the zone's DNSSEC
keys on a schedule, according to the keys' timing metadata
(see dnssec-keygen(8) and
dnssec-settime(8)). The command
rndc sign
zonename causes
named to load keys from the key
repository and sign the zone with all keys that are
active.
rndc loadkeys
zonename causes
named to load keys from the key
repository and schedule key maintenance events to occur
in the future, but it does not sign the full zone
immediately. Note: once keys have been loaded for a
zone the first time, the repository is searched
for changes periodically, regardless of whether
rndc loadkeys is used. The recheck
interval is defined by
dnssec-loadkeys-interval.)
The default setting is auto-dnssec off.
- dnssec-enable
This indicates whether DNSSEC-related resource
records are to be returned by named.
If set to no,
named does not return DNSSEC-related
resource records unless specifically queried for.
The default is yes.
- dnssec-validation
-
This option enables DNSSEC validation in named.
Note that dnssec-enable also needs to be
set to yes to be effective.
If set to no, DNSSEC validation
is disabled.
If set to auto, DNSSEC validation
is enabled and a default trust anchor for the DNS root
zone is used. If set to yes,
DNSSEC validation is enabled, but a trust anchor must be
manually configured using a trusted-keys
or managed-keys statement. The default
is yes.
The default root trust anchor is stored in the file
bind.keys.
named loads that key at
startup if dnssec-validation is
set to auto. A copy of the file is
installed along with BIND 9, and is current as of the
release date. If the root key expires, a new copy of
bind.keys can be downloaded
from https://www.isc.org/bind-keys.
(To prevent problems if bind.keys is
not found, the current trust anchor is also compiled in
to named. Relying on this is not
recommended, however, as it requires named
to be recompiled with a new key when the root key expires.)
Note
named loads only
the root key from bind.keys.
The file cannot be used to store keys for other zones.
The root key in bind.keys is ignored
if dnssec-validation auto is not in
use.
Whenever the resolver sends out queries to an
EDNS-compliant server, it always sets the DO bit
indicating it can support DNSSEC responses, even if
dnssec-validation is off.
- dnssec-accept-expired
This accepts expired signatures when verifying DNSSEC signatures.
The default is no.
Setting this option to yes
leaves named vulnerable to
replay attacks.
- querylog
-
Query logging provides a complete log of all incoming
queries and all query errors. This provides more insight
into the server's activity, but with a cost to
performance which may be significant on heavily loaded
servers.
The querylog option specifies
whether query logging should be active when
named first starts.
If querylog is not specified, then
query logging is determined by the presence of the
logging category queries.
Query logging can also be activated at runtime using the
command rndc querylog on, or
deactivated with rndc querylog off.
- check-names
-
This option is used to restrict the character set and syntax
of
certain domain names in zone files and/or DNS responses
received
from the network. The default varies according to usage
area. For primary zones (i.e.,
type master),
the default is fail.
For secondary zones (type slave), the
default is warn.
For answers received from the network (response),
the default is ignore.
The rules for legal hostnames and mail domains are derived
from RFC 952 and RFC 821 as modified by RFC 1123.
check-names
applies to the owner names of A, AAAA, and MX records.
It also applies to the domain names in the RDATA of NS, SOA,
MX, and SRV records.
It further applies to the RDATA of PTR records where the owner
name indicates that it is a reverse lookup of a hostname
(the owner name ends in IN-ADDR.ARPA, IP6.ARPA, or IP6.INT).
- check-dup-records
This checks primary zones for records that are treated as different
by DNSSEC but are semantically equal in plain DNS. The
default is to warn. Other possible
values are fail and
ignore.
- check-mx
This checks whether the MX record appears to refer to a IP address.
The default is to warn. Other possible
values are fail and
ignore.
- check-wildcard
This option is used to check for non-terminal wildcards.
The use of non-terminal wildcards is almost always as a
result of a failure
to understand the wildcard matching algorithm (RFC 1034).
This option
affects primary zones. The default (yes) is to check
for non-terminal wildcards and issue a warning.
- check-integrity
-
This performs post-load zone integrity checks on primary
zones. It checks that MX and SRV records refer
to address (A or AAAA) records and that glue
address records exist for delegated zones. For
MX and SRV records, only in-zone hostnames are
checked (for out-of-zone hostnames, use
named-checkzone).
For NS records, only names below top-of-zone are
checked (for out-of-zone names and glue consistency
checks, use named-checkzone).
The default is yes.
The use of the SPF record to publish Sender
Policy Framework is deprecated, as the migration
from using TXT records to SPF records was abandoned.
Enabling this option also checks that a TXT Sender
Policy Framework record exists (starts with "v=spf1")
if there is an SPF record. Warnings are emitted if the
TXT record does not exist; they can be suppressed with
check-spf.
- check-mx-cname
If check-integrity is set, then
fail, warn, or ignore MX records that refer
to CNAMES. The default is to warn.
- check-srv-cname
If check-integrity is set, then
fail, warn, or ignore SRV records that refer
to CNAMES. The default is to warn.
- check-sibling
When performing integrity checks, also check that
sibling glue exists. The default is yes.
- check-spf
If check-integrity is set,
check that there is a TXT Sender Policy Framework
record present (starts with "v=spf1") if there is an
SPF record present. The default is
warn.
- zero-no-soa-ttl
If yes, when returning authoritative negative responses to
SOA queries, set the TTL of the SOA record returned in
the authority section to zero.
The default is yes.
- zero-no-soa-ttl-cache
If yes, when caching a negative response to an SOA query
set the TTL to zero.
The default is no.
- update-check-ksk
-
When set to the default value of yes,
check the KSK bit in each key to determine how the key
should be used when generating RRSIGs for a secure zone.
Ordinarily, zone-signing keys (that is, keys without the
KSK bit set) are used to sign the entire zone, while
key-signing keys (keys with the KSK bit set) are only
used to sign the DNSKEY RRset at the zone apex.
However, if this option is set to no,
then the KSK bit is ignored; KSKs are treated as if they
were ZSKs and are used to sign the entire zone. This is
similar to the dnssec-signzone -z
command-line option.
When this option is set to yes, there
must be at least two active keys for every algorithm
represented in the DNSKEY RRset: at least one KSK and one
ZSK per algorithm. If there is any algorithm for which
this requirement is not met, this option is ignored
for that algorithm.
- dnssec-dnskey-kskonly
-
When this option and update-check-ksk
are both set to yes, only key-signing
keys (that is, keys with the KSK bit set) are used
to sign the DNSKEY RRset at the zone apex. Zone-signing
keys (keys without the KSK bit set) are used to sign
the remainder of the zone, but not the DNSKEY RRset.
This is similar to the
dnssec-signzone -x command-line option.
The default is no. If
update-check-ksk is set to
no, this option is ignored.
- try-tcp-refresh
If yes, try to refresh the zone using TCP if UDP queries fail.
The default is
yes.
- dnssec-secure-to-insecure
-
This allows a dynamic zone to transition from secure to
insecure (i.e., signed to unsigned) by deleting all
of the DNSKEY records. The default is no.
If set to yes, and if the DNSKEY RRset
at the zone apex is deleted, all RRSIG and NSEC records
are removed from the zone as well.
If the zone uses NSEC3, it is also necessary to
delete the NSEC3PARAM RRset from the zone apex; this
causes the removal of all corresponding NSEC3 records.
(It is expected that this requirement will be eliminated
in a future release.)
Note that if a zone has been configured with
auto-dnssec maintain and the
private keys remain accessible in the key repository,
then the zone will be automatically signed again the
next time named is started.
The forwarding facility can be used to create a large site-wide
cache on a few servers, reducing traffic over links to external
name servers. It can also be used to allow queries by servers that
do not have direct access to the Internet, but wish to look up
exterior
names anyway. Forwarding occurs only on those queries for which
the server is not authoritative and does not have the answer in
its cache.
- forward
This option is only meaningful if the
forwarders list is not empty. A value of first is
the default and causes the server to query the forwarders
first;
if that does not answer the question, the server then
looks for
the answer itself. If only is
specified, the
server only queries the forwarders.
- forwarders
This specifies a list of IP addresses to which queries are
forwarded. The default is the empty list (no forwarding).
Each address in the list can be associated with an optional
port number and/or DSCP value, and a default port number and
DSCP value can be set for the entire list.
Forwarding can also be configured on a per-domain basis, allowing
for the global forwarding options to be overridden in a variety
of ways. Particular domains can be set to use different
forwarders,
or have a different forward only/first behavior,
or not forward at all; see the section called “zone
Statement Grammar”.
Dual-stack servers are used as servers of last resort, to work
around
problems in reachability due the lack of support for either IPv4
or IPv6
on the host machine.
- dual-stack-servers
This specifies host names or addresses of machines with access to
both IPv4 and IPv6 transports. If a hostname is used, the
server must be able
to resolve the name using only the transport it has. If the
machine is dual-stacked, the dual-stack-servers parameter has no effect unless
access to a transport has been disabled on the command line
(e.g., named -4).
Access to the server can be restricted based on the IP address
of the requesting system. See the section called “Address Match Lists” for
details on how to specify IP address lists.
- allow-notify
This ACL specifies which hosts are allowed to
notify this secondary server of zone changes in addition
to the zone primaries.
allow-notify may also be
specified in the
zone statement, in which case
it overrides the
options allow-notify
statement. It is only meaningful
for a secondary zone. If not specified, the default is to
process notify messages
only from a zone's primary.
- allow-query
-
This specifies which hosts are allowed to ask ordinary
DNS questions. allow-query may
also be specified in the zone
statement, in which case it overrides the
options allow-query statement.
If not specified, the default is to allow queries
from all hosts.
Note
allow-query-cache is
used to specify access to the cache.
- allow-query-on
-
This specifies which local addresses can accept ordinary
DNS questions. This makes it possible, for instance,
to allow queries on internal-facing interfaces but
disallow them on external-facing ones, without
necessarily knowing the internal network's addresses.
Note that allow-query-on is only
checked for queries that are permitted by
allow-query. A query must be
allowed by both ACLs, or it is refused.
allow-query-on may
also be specified in the zone
statement, in which case it overrides the
options allow-query-on statement.
If not specified, the default is to allow queries
on all addresses.
Note
allow-query-cache is
used to specify access to the cache.
- allow-query-cache
This specifies which hosts are allowed to get answers
from the cache. If allow-query-cache
is not set, BIND checks to see if the following parameters
are set, in order: allow-recursion and
allow-query
(unless recursion no; is
set, in which case none; is used).
If neither of those parameters is set, the default (localnets;
localhost;) is used.
- allow-query-cache-on
This specifies which local addresses can send answers
from the cache. If not specified, the default is
to allow cache queries on any address,
localnets, and
localhost.
- allow-recursion
This specifies which hosts are allowed to make recursive
queries through this server. BIND checks to see if the
following parameters are set, in order:
allow-recursion, allow-query-cache,
and allow-query.
If none of those parameters are set, the default
(localnets;
localhost;) is used.
- allow-recursion-on
This specifies which local addresses can accept recursive
queries. If not specified, the default is to allow
recursive queries on all addresses.
- allow-update
This specifies which hosts are allowed to
submit Dynamic DNS updates for primary zones. The default is
to deny
updates from all hosts. Note that allowing updates based
on the requestor's IP address is insecure; see
the section called “Dynamic Update Security” for details.
- allow-update-forwarding
-
This specifies which hosts are allowed to
submit Dynamic DNS updates to secondary zones to be forwarded to
the
primary. The default is { none; },
which
means that no update forwarding is performed. To
enable
update forwarding, specify
allow-update-forwarding { any; };.
Specifying values other than { none; } or
{ any; } is usually
counterproductive;
the responsibility for update access control should rest
with the
primary server, not the secondaries.
Note that enabling the update forwarding feature on a secondary
server
may expose primary servers to attacks if they rely on insecure
IP-address-based
access control; see the section called “Dynamic Update Security”
for more details.
- allow-v6-synthesis
This option was introduced for the smooth transition from
AAAA
to A6 and from "nibble labels" to binary labels.
However, since both A6 and binary labels were then
deprecated,
this option was also deprecated.
It is now ignored with some warning messages.
- allow-transfer
This specifies which hosts are allowed to
receive zone transfers from the server. allow-transfer may
also be specified in the zone
statement, in which
case it overrides the options allow-transfer statement.
If not specified, the default is to allow transfers to all
hosts.
- blackhole
This specifies a list of addresses which the
server does accept queries from or use to resolve a
query. Queries
from these addresses are not responded to. The default
is none.
- filter-aaaa
This specifies a list of addresses to which
filter-aaaa-on-v4
and filter-aaaa-on-v6
apply. The default is any.
- keep-response-order
This specifies a list of addresses to which the server
sends responses to TCP queries, in the same order
in which they were received. This disables the
processing of TCP queries in parallel. The default
is none.
- no-case-compress
-
This specifies a list of addresses which require responses
to use case-insensitive compression. This ACL can be
used when named needs to work with
clients that do not comply with the requirement in RFC
1034 to use case-insensitive name comparisons when
checking for matching domain names.
If left undefined, the ACL defaults to
none: case-insensitive compression
is used for all clients. If the ACL is defined and
matches a client, case is ignored when
compressing domain names in DNS responses sent to that
client.
This can result in slightly smaller responses; if
a response contains the names "example.com" and
"example.COM", case-insensitive compression treats
the second one as a duplicate. It also ensures
that the case of the query name exactly matches the
case of the owner names of returned records, rather
than matches the case of the records entered in
the zone file. This allows responses to exactly
match the query, which is required by some clients
due to incorrect use of case-sensitive comparisons.
Case-insensitive compression is always
used in AXFR and IXFR responses, regardless of whether
the client matches this ACL.
There are circumstances in which named
does not preserve the case of owner names of records:
if a zone file defines records of different types with
the same name, but the capitalization of the name is
different (e.g., "www.example.com/A" and
"WWW.EXAMPLE.COM/AAAA"), then all responses for that
name use the first version
of the name that was used in the zone file. This
limitation may be addressed in a future release. However,
domain names specified in the rdata of resource records
(i.e., records of type NS, MX, CNAME, etc.) always
have their case preserved unless the client matches this
ACL.
- resolver-query-timeout
This is the amount of time in seconds that the
resolver spends attempting to resolve a recursive
query before failing. The default and minimum
is 10 and the maximum is
30. Setting it to
0 results in the default
being used.
The interfaces and ports that the server answers queries
from may be specified using the listen-on option. listen-on takes
an optional port and an address_match_list
of IPv4 addresses. (IPv6 addresses are ignored, with a
logged warning.)
The server listens on all interfaces allowed by the address
match list. If a port is not specified, port 53 is used.
Multiple listen-on statements are
allowed.
For example:
listen-on { 5.6.7.8; };
listen-on port 1234 { !1.2.3.4; 1.2/16; };
enables the name server on port 53 for the IP address
5.6.7.8, and on port 1234 of an address on the machine in net
1.2 that is not 1.2.3.4.
If no listen-on is specified, the
server listens on port 53 on all IPv4 interfaces.
The listen-on-v6 option is used to
specify the interfaces and the ports on which the server
listens for incoming queries sent using IPv6. If not specified,
the server listens on port 53 on all IPv6 interfaces.
When
{ any; }
is
specified
as the address_match_list for the
listen-on-v6 option,
the server does not bind a separate socket to each IPv6 interface
address as it does for IPv4, if the operating system has enough API
support for IPv6 (specifically, if it conforms to RFC 3493 and RFC
3542).
Instead, it listens on the IPv6 wildcard address.
If the system only has incomplete API support for IPv6, however,
the behavior is the same as that for IPv4.
A list of particular IPv6 addresses can also be specified, in
which case
the server listens on a separate socket for each specified
address,
regardless of whether the desired API is supported by the system.
IPv4 addresses specified in listen-on-v6
are ignored, with a logged warning.
Multiple listen-on-v6 options can
be used.
For example:
listen-on-v6 { any; };
listen-on-v6 port 1234 { !2001:db8::/32; any; };
enables the name server on port 53 for any IPv6 addresses
(with a single wildcard socket),
and on port 1234 of IPv6 addresses that are not in the prefix
2001:db8::/32 (with separate sockets for each matched address).
To instruct the server not to listen on any IPv6 address, use:
listen-on-v6 { none; };
If the server does not know the answer to a question, it
queries other name servers. query-source specifies
the address and port used for such queries. For queries sent over
IPv6, there is a separate query-source-v6 option.
If address is * (asterisk) or is omitted,
a wildcard IP address (INADDR_ANY)
is used.
If port is * or is omitted,
a random port number from a pre-configured
range is picked up and used for each query.
The port range(s) is specified in
the use-v4-udp-ports (for IPv4)
and use-v6-udp-ports (for IPv6)
options, excluding the ranges specified in
the avoid-v4-udp-ports
and avoid-v6-udp-ports options, respectively.
The defaults of the query-source and
query-source-v6 options
are:
query-source address * port *;
query-source-v6 address * port *;
If use-v4-udp-ports or
use-v6-udp-ports is unspecified,
named checks whether the operating
system provides a programming interface to retrieve the
system's default range for ephemeral ports.
If such an interface is available,
named uses the corresponding system
default range; otherwise, it uses its own defaults:
use-v4-udp-ports { range 1024 65535; };
use-v6-udp-ports { range 1024 65535; };
Note: make sure the ranges are sufficiently large for
security. A desirable size depends on several parameters,
but we generally recommend it contain at least 16384 ports
(14 bits of entropy).
Note also that the system's default range when used may be
too small for this purpose, and that the range may even be
changed while named is running; the new
range is automatically applied when named
is reloaded.
Explicit
configuration of use-v4-udp-ports and
use-v6-udp-ports is encouraged, so that the
ranges are sufficiently large and are reasonably
independent from the ranges used by other applications.
Note: the operational configuration
where named runs may prohibit the use
of some ports. For example, Unix systems do not allow
named, if run without root privilege,
to use ports less than 1024.
If such ports are included in the specified (or detected)
set of query ports, the corresponding query attempts will
fail, resulting in resolution failures or delay.
It is therefore important to configure the set of ports
that can be safely used in the expected operational environment.
The defaults of the avoid-v4-udp-ports and
avoid-v6-udp-ports options
are:
avoid-v4-udp-ports {};
avoid-v6-udp-ports {};
Note: BIND 9.5.0 introduced
the use-queryport-pool
option to support a pool of such random ports, but this
option is now obsolete because reusing the same ports in
the pool may not be sufficiently secure.
For the same reason, it is generally strongly discouraged to
specify a particular port for the
query-source or
query-source-v6 options;
it implicitly disables the use of randomized port numbers.
- use-queryport-pool
This option is obsolete.
- queryport-pool-ports
This option is obsolete.
- queryport-pool-updateinterval
This option is obsolete.
Note
The address specified in the query-source option
is used for both UDP and TCP queries, but the port applies only
to UDP queries. TCP queries always use a random
unprivileged port.
Note
Solaris 2.5.1 and earlier does not support setting the source
address for TCP sockets.
Note
See also transfer-source and
notify-source.
BIND has mechanisms in place to
facilitate zone transfers
and set limits on the amount of load that transfers place on the
system. The following options apply to zone transfers.
- also-notify
-
This option defines a global list of IP addresses of name servers
that are also sent NOTIFY messages whenever a fresh copy of
the
zone is loaded, in addition to the servers listed in the
zone's NS records.
This helps to ensure that copies of the zones
quickly converge on stealth servers.
Optionally, a port may be specified with each
also-notify address to send
the notify messages to a port other than the
default of 53.
An optional TSIG key can also be specified with each
address to cause the notify messages to be signed; this
can be useful when sending notifies to multiple views.
In place of explicit addresses, one or more named
masters lists can be used.
If an also-notify list
is given in a zone statement,
it overrides
the options also-notify
statement. When a zone notify
statement
is set to no, the IP
addresses in the global also-notify list are
not sent NOTIFY messages for that zone. The default is
the empty
list (no global notification list).
- max-transfer-time-in
Inbound zone transfers running longer than
this many minutes are terminated. The default is 120
minutes
(2 hours). The maximum value is 28 days (40320 minutes).
- max-transfer-idle-in
Inbound zone transfers making no progress
in this many minutes are terminated. The default is 60
minutes
(1 hour). The maximum value is 28 days (40320 minutes).
- max-transfer-time-out
Outbound zone transfers running longer than
this many minutes are terminated. The default is 120
minutes
(2 hours). The maximum value is 28 days (40320 minutes).
- max-transfer-idle-out
Outbound zone transfers making no progress
in this many minutes are terminated. The default is 60
minutes (1
hour). The maximum value is 28 days (40320 minutes).
- notify-rate
This specifies the rate at which NOTIFY requests are sent
during normal zone maintenance operations. (NOTIFY
requests due to initial zone loading are subject
to a separate rate limit; see below.) The default is
20 per second.
The lowest possible rate is one per second; when set
to zero, it is silently raised to one.
- startup-notify-rate
This is the rate at which NOTIFY requests are sent
when the name server is first starting up, or when
zones have been newly added to the name server.
The default is 20 per second.
The lowest possible rate is one per second; when set
to zero, it is silently raised to one.
- serial-query-rate
Secondary servers periodically query primary
servers to find out if zone serial numbers have
changed. Each such query uses a minute amount of
the secondary server's network bandwidth. To limit
the amount of bandwidth used, BIND 9 limits the
rate at which queries are sent. The value of the
serial-query-rate option, an
integer, is the maximum number of queries sent
per second. The default is 20 per second.
The lowest possible rate is one per second; when set
to zero, it is silently raised to one.
- serial-queries
BIND 9 does not limit the number of outstanding
serial queries and ignores the serial-queries option.
Instead, it limits the rate at which the queries are sent
as defined using the serial-query-rate option.
- transfer-format
Zone transfers can be sent using two different formats,
one-answer and
many-answers.
The transfer-format option is used
on the primary server to determine which format it sends.
one-answer uses one DNS message per
resource record transferred.
many-answers packs as many resource
records as possible into one message.
many-answers is more efficient; the default is many-answers.
The many-answers format is also supported by
recent Microsoft Windows name servers.
transfer-format may be overridden on a
per-server basis by using the server
statement.
- transfer-message-size
-
This is an upper bound on the uncompressed size of DNS
messages used in zone transfers over TCP. If a message
grows larger than this size, additional messages are
used to complete the zone transfer. (Note, however,
that this is a hint, not a hard limit; if a message
contains a single resource record whose RDATA does not
fit within the size limit, a larger message will be
permitted so the record can be transferred.)
Valid values are between 512 and 65535 octets; any
values outside that range are adjusted to the nearest
value within it. The default is 20480,
which was selected to improve message compression;
most DNS messages of this size will compress to less
than 16536 bytes. Larger messages cannot be compressed
as effectively, because 16536 is the largest permissible
compression offset pointer in a DNS message.
This option is mainly intended for server testing;
there is rarely any benefit in setting a value other
than the default.
- transfers-in
This is the maximum number of inbound zone transfers
that can run concurrently. The default value is 10.
Increasing transfers-in may
speed up the convergence
of secondary zones, but it also may increase the load on the
local system.
- transfers-out
This is the maximum number of outbound zone transfers
that can run concurrently. Zone transfer requests in
excess
of the limit are refused. The default value is 10.
- transfers-per-ns
This is the maximum number of inbound zone transfers
that can concurrently transfer from a given remote
name server.
The default value is 2.
Increasing transfers-per-ns
may
speed up the convergence of secondary zones, but it also may
increase
the load on the remote name server. transfers-per-ns may
be overridden on a per-server basis by using the transfers phrase
of the server statement.
- transfer-source
-
transfer-source
determines which local address is bound to IPv4
TCP connections used to fetch zones transferred
inbound by the server. It also determines the
source IPv4 address, and optionally the UDP port,
used for the refresh queries and forwarded dynamic
updates. If not set, it defaults to a
system-controlled value which is usually the address
of the interface "closest to" the remote end. This
address must appear in the remote end's
allow-transfer option for the
zone being transferred, if one is specified. This
statement sets the
transfer-source for all zones,
but can be overridden on a per-view or per-zone
basis by including a
transfer-source statement within
the view or
zone block in the configuration
file.
Note
Solaris 2.5.1 and earlier does not support setting the
source address for TCP sockets.
- transfer-source-v6
This option is the same as transfer-source,
except zone transfers are performed using IPv6.
- alt-transfer-source
-
This indicates an alternate transfer source if the one listed in
transfer-source fails and
use-alt-transfer-source is
set.
Note
To avoid using the alternate transfer source,
set use-alt-transfer-source
appropriately and do not depend upon
getting an answer back to the first refresh
query.
- alt-transfer-source-v6
This indicates an alternate transfer source if the one listed in
transfer-source-v6 fails and
use-alt-transfer-source is
set.
- use-alt-transfer-source
This indicates whether the alternate transfer sources should be used. If views are
specified, this defaults to no;
otherwise, it defaults to
yes.
- notify-source
-
notify-source
determines which local source address, and
optionally UDP port, is used to send NOTIFY
messages. This address must appear in the secondary
server's masters zone clause or
in an allow-notify clause. This
statement sets the notify-source
for all zones, but can be overridden on a per-zone or
per-view basis by including a
notify-source statement within
the zone or
view block in the configuration
file.
Note
Solaris 2.5.1 and earlier does not support setting the
source address for TCP sockets.
- notify-source-v6
This option acts like notify-source,
but applies to notify messages sent to IPv6 addresses.
use-v4-udp-ports,
avoid-v4-udp-ports,
use-v6-udp-ports, and
avoid-v6-udp-ports
specify a list of IPv4 and IPv6 UDP ports that are
or are not used as source ports for UDP messages.
See the section called “Query Address” about how the
available ports are determined.
For example, with the following configuration:
use-v6-udp-ports { range 32768 65535; };
avoid-v6-udp-ports { 40000; range 50000 60000; };
UDP ports of IPv6 messages sent
from named are in one
of the following ranges: 32768 to 39999, 40001 to 49999,
and 60001 to 65535.
avoid-v4-udp-ports and
avoid-v6-udp-ports can be used
to prevent named from choosing as its random source port a
port that is blocked by a firewall or a port that is
used by other applications;
if a query went out with a source port blocked by a
firewall, the
answer would not pass through the firewall and the name server would
have to query again.
Note: the desired range can also be represented only with
use-v4-udp-ports and
use-v6-udp-ports, and the
avoid- options are redundant in that
sense; they are provided for backward compatibility and
to possibly simplify the port specification.
Operating System Resource Limits
The server's usage of many system resources can be limited.
Scaled values are allowed when specifying resource limits. For
example, 1G can be used instead of
1073741824 to specify a limit of
one
gigabyte. unlimited requests
unlimited use, or the
maximum available amount. default
uses the limit
that was in force when the server was started. See the description
of size_spec in the section called “Configuration File Elements”.
The following options set operating system resource limits for
the name server process. Some operating systems do not support
some or
any of the limits; on such systems, a warning is issued if
an
unsupported limit is used.
- coresize
This sets the maximum size of a core dump. The default
is default.
- datasize
This sets the maximum amount of data memory the server
may use. The default is default.
This is a hard limit on server memory usage;
if the server attempts to allocate memory in excess of this
limit, the allocation will fail, which may in turn leave
the server unable to perform DNS service. Therefore,
this option is rarely useful as a way to limit the
amount of memory used by the server, but it can be used
to raise an operating system data size limit that is
too small by default. To limit the amount
of memory used by the server, use the
max-cache-size and
recursive-clients
options instead.
- files
This sets the maximum number of files the server
may have open concurrently. The default is unlimited.
- stacksize
This sets the maximum amount of stack memory the server
may use. The default is default.
The following options set limits on the server's
resource consumption that are enforced internally by the
server rather than by the operating system.
- max-ixfr-log-size
This option is obsolete; it is accepted
and ignored for BIND 8 compatibility. The option
max-journal-size performs a
similar function in BIND 9.
- max-journal-size
This sets a maximum size for each journal file
(see the section called “The Journal File”). When the journal file
approaches
the specified size, some of the oldest transactions in the
journal
are automatically removed. The largest permitted
value is 2 gigabytes. The default is
unlimited, which also
means 2 gigabytes.
This option may also be set on a per-zone basis.
- max-records
This sets the maximum number of records permitted in a zone.
The default is zero, which means the maximum is unlimited.
- host-statistics-max
In BIND 8, this specified the maximum number of host statistics
entries to be kept.
It is not implemented in BIND 9.
- recursive-clients
-
This sets the maximum number (a "hard quota") of simultaneous
recursive lookups the server performs on behalf
of clients. The default is
1000. Because each recursing
client uses a fair
bit of memory (on the order of 20 kilobytes), the
value of the
recursive-clients option may
have to be decreased on hosts with limited memory.
recursive-clients defines a "hard
quota" limit for pending recursive clients; when more
clients than this are pending, new incoming requests
are not accepted, and for each incoming request
a previous pending request is dropped.
A "soft quota" is also set. When this lower
quota is exceeded, incoming requests are accepted, but
for each one, a pending request is dropped.
If recursive-clients is greater than
1000, the soft quota is set to
recursive-clients minus 100;
otherwise it is set to 90% of
recursive-clients.
- tcp-clients
This is the maximum number of simultaneous client TCP
connections that the server accepts.
The default is 150.
-
clients-per-query, max-clients-per-query
-
These set the
initial value (minimum) and maximum number of recursive
simultaneous clients for any given query
(<qname,qtype,qclass>) that the server accepts
before dropping additional clients. named attempts to
self-tune this value and changes are logged. The
default values are 10 and 100.
This value should reflect how many queries come in for
a given name in the time it takes to resolve that name.
If the number of queries exceeds this value, named
assumes that it is dealing with a non-responsive zone
and drops additional queries. If it gets a response
after dropping queries, it raises the estimate. The
estimate is then lowered in 20 minutes if it has
remained unchanged.
If clients-per-query is set to zero,
there is no limit on the number of clients per query
and no queries are dropped.
If max-clients-per-query is set to zero,
there is no upper bound other than imposed by
recursive-clients.
-
fetches-per-zone
-
This sets the maximum number of simultaneous iterative
queries to any one domain that the server
permits before blocking new queries for data
in or beneath that zone.
This value should reflect how many fetches would
normally be sent to any one zone in the time it
would take to resolve them. It should be smaller
than recursive-clients.
When many clients simultaneously query for the
same name and type, the clients are all attached
to the same fetch, up to the
max-clients-per-query limit,
and only one iterative query is sent.
However, when clients are simultaneously
querying for different names
or types, multiple queries are sent and
max-clients-per-query is not
effective as a limit.
Optionally, this value may be followed by the keyword
drop or fail,
indicating whether queries which exceed the fetch
quota for a zone are dropped with no response,
or answered with SERVFAIL. The default is
drop.
If fetches-per-zone is set to zero,
there is no limit on the number of fetches per query
and no queries are dropped. The default is zero.
The current list of active fetches can be dumped by
running rndc recursing. The list
includes the number of active fetches for each
domain and the number of queries that have been
passed or dropped as a result of the
fetches-per-zone limit. (Note:
these counters are not cumulative over time; whenever
the number of active fetches for a domain drops to
zero, the counter for that domain is deleted, and the
next time a fetch is sent to that domain, it is
recreated with the counters set to zero.)
-
fetches-per-server
-
This sets the maximum number of simultaneous iterative
queries that the server allows to be sent to
a single upstream name server before blocking
additional queries.
This value should reflect how many fetches would
normally be sent to any one server in the time it
would take to resolve them. It should be smaller
than recursive-clients.
Optionally, this value may be followed by the keyword
drop or fail,
indicating whether queries are dropped with no
response or answered with SERVFAIL, when all of the
servers authoritative for a zone are found to have
exceeded the per-server quota. The default is
fail.
If fetches-per-server is set to zero,
there is no limit on the number of fetches per query
and no queries are dropped. The default is zero.
The fetches-per-server quota is
dynamically adjusted in response to detected
congestion. As queries are sent to a server
and are either answered or time out, an
exponentially weighted moving average is calculated
of the ratio of timeouts to responses. If the
current average timeout ratio rises above a "high"
threshold, then fetches-per-server
is reduced for that server. If the timeout ratio
drops below a "low" threshold, then
fetches-per-server is increased.
The fetch-quota-params options
can be used to adjust the parameters for this
calculation.
- fetch-quota-params
-
This sets the parameters to use for dynamic resizing of
the fetches-per-server quota in
response to detected congestion.
The first argument is an integer value indicating
how frequently to recalculate the moving average
of the ratio of timeouts to responses for each
server. The default is 100, meaning that BIND recalculates
the average ratio after every 100 queries have either
been answered or timed out.
The remaining three arguments represent the "low"
threshold (defaulting to a timeout ratio of 0.1),
the "high" threshold (defaulting to a timeout
ratio of 0.3), and the discount rate for
the moving average (defaulting to 0.7).
A higher discount rate causes recent events to
weigh more heavily when calculating the moving
average; a lower discount rate causes past
events to weigh more heavily, smoothing out
short-term blips in the timeout ratio.
These arguments are all fixed-point numbers with
precision of 1/100; at most two places after
the decimal point are significant.
- reserved-sockets
-
This sets the number of file descriptors reserved for TCP, stdio,
etc. This needs to be big enough to cover the number of
interfaces named listens on plus
tcp-clients, as well as
to provide room for outgoing TCP queries and incoming zone
transfers. The default is 512.
The minimum value is 128 and the
maximum value is 128 fewer than
maxsockets (-S). This option may be removed in the future.
This option has little effect on Windows.
- max-cache-size
This sets the maximum amount of memory to use for the
server's cache, in bytes or percentage of total physical memory.
When the amount of data in the cache
reaches this limit, the server causes records to
expire prematurely, following an LRU-based strategy, so
that the limit is not exceeded.
The keyword unlimited,
or the value 0, places no limit on the cache size;
records are purged from the cache only when their
TTLs expire.
Any positive values less than 2MB are ignored
and reset to 2MB.
In a server with multiple views, the limit applies
separately to the cache of each view.
The default is 90%.
On systems where detection of the amount of physical
memory is not supported, values represented as a percentage
fall back to unlimited.
Note that the detection of physical memory is done only
once at startup, so named does not
adjust the cache size if the amount of physical memory
is changed during runtime.
- tcp-listen-queue
This sets the listen-queue depth. The default and minimum is 10.
If the kernel supports the accept filter "dataready", this
also controls how
many TCP connections are queued in kernel space
waiting for
some data before being passed to accept. Non-zero values
less than 10 are silently raised. A value of 0 may also
be used; on most platforms this sets the listen-queue
length to a system-defined default value.
- cleaning-interval
This interval is effectively obsolete. Previously,
the server removed expired resource records
from the cache every cleaning-interval minutes.
BIND 9 now manages cache
memory in a more sophisticated manner and does not
rely on periodic cleaning anymore.
Specifying this option therefore has no effect on
the server's behavior.
- heartbeat-interval
The server performs zone maintenance tasks
for all zones marked as dialup whenever this
interval expires. The default is 60 minutes. Reasonable
values are up
to 1 day (1440 minutes). The maximum value is 28 days
(40320 minutes).
If set to 0, no zone maintenance for these zones occurs.
- interface-interval
The server scans the network interface list
every interface-interval
minutes. The default
is 60 minutes; the maximum value is 28 days (40320 minutes).
If set to 0, interface scanning only occurs when
the configuration file is loaded, or when
automatic-interface-scan is enabled
and supported by the operating system. After the scan, the
server begins listening for queries on any newly
discovered interfaces (provided they are allowed by the
listen-on configuration), and
stops listening on interfaces that have gone away.
- statistics-interval
-
Name server statistics are logged
every statistics-interval
minutes. The default is
60, and the maximum value is 28 days (40320 minutes).
If set to 0, no statistics are logged.
Note
This option is not implemented in
BIND 9.
- topology
In BIND 8, this option indicated network topology
so that preferential treatment could be given to
the topologically closest name servers when sending
queries. It is not implemented in BIND 9.
The response to a DNS query may consist of multiple resource
records (RRs) forming a resource record set (RRset). The name
server normally returns the RRs within the RRset in an
indeterminate order (but see the rrset-order
statement in the section called “RRset Ordering”). The client
resolver code should rearrange the RRs as appropriate: that is,
using any addresses on the local net in preference to other
addresses. However, not all resolvers can do this or are
correctly configured. When a client is using a local server,
the sorting can be performed in the server, based on the
client's address. This only requires configuring the name
servers, not all the clients.
The sortlist statement (see below) takes an
address_match_list and interprets it in a
special way. Each top-level statement in the
sortlist must itself be an explicit
address_match_list with one or two elements.
The first element (which may be an IP address, an IP prefix, an
ACL name, or a nested address_match_list) of
each top-level list is checked against the source address of
the query until a match is found. When the addresses in the
first element overlap, the first rule to match is selected.
Once the source address of the query has been matched, if the
top-level statement contains only one element, the actual
primitive element that matched the source address is used to
select the address in the response to move to the beginning of
the response. If the statement is a list of two elements, then
the second element is interpreted as a topology preference
list. Each top-level element is assigned a distance, and the
address in the response with the minimum distance is moved to
the beginning of the response.
In the following example, any queries received from any of the
addresses of the host itself get responses preferring
addresses on any of the locally connected networks. Next most
preferred are addresses on the 192.168.1/24 network, and after
that either the 192.168.2/24 or 192.168.3/24 network, with no
preference shown between these two networks. Queries received
from a host on the 192.168.1/24 network prefer other
addresses on that network to the 192.168.2/24 and 192.168.3/24
networks. Queries received from a host on the 192.168.4/24 or
the 192.168.5/24 network only prefer other addresses on
their directly connected networks.
sortlist {
// IF the local host
// THEN first fit on the following nets
{ localhost;
{ localnets;
192.168.1/24;
{ 192.168.2/24; 192.168.3/24; }; }; };
// IF on class C 192.168.1 THEN use .1, or .2 or .3
{ 192.168.1/24;
{ 192.168.1/24;
{ 192.168.2/24; 192.168.3/24; }; }; };
// IF on class C 192.168.2 THEN use .2, or .1 or .3
{ 192.168.2/24;
{ 192.168.2/24;
{ 192.168.1/24; 192.168.3/24; }; }; };
// IF on class C 192.168.3 THEN use .3, or .1 or .2
{ 192.168.3/24;
{ 192.168.3/24;
{ 192.168.1/24; 192.168.2/24; }; }; };
// IF .4 or .5 THEN prefer that net
{ { 192.168.4/24; 192.168.5/24; };
};
};
The following example illustrates reasonable behavior for the
local host and hosts on directly connected networks. Responses sent to queries from
the local host favor any of the directly connected
networks. Responses sent to queries from any other hosts on a
directly connected network prefer addresses on that same
network. Responses to other queries are not sorted.
sortlist {
{ localhost; localnets; };
{ localnets; };
};
Note
While alternating the order of records in a DNS response between
subsequent queries is a known load distribution technique, certain
caveats apply (mostly stemming from caching) which usually make it
a suboptimal choice for load balancing purposes when used on its
own.
The rrset-order statement permits configuration
of the ordering of the records in a multiple-record response. See
also: the section called “The sortlist Statement”.
Each rule in an rrset-order statement is defined
as follows:
[class <class_name>]
[type <type_name>]
[name "<domain_name>"]
order <ordering>
The default qualifiers for each rule are:
-
If no class is specified, the default is
ANY.
-
If no type is specified, the default is
ANY.
-
If no name is specified, the default is
* (asterisk).
<domain_name> only matches the name
itself, not any of its subdomains. To make a rule match all
subdomains of a given name, a wildcard name
(*.<domain_name>) must be used.
Note that *.<domain_name> does
not match
<domain_name> itself; to specify
RRset ordering for a name and all of its subdomains, two separate
rules must be defined: one for
<domain_name> and one for
*.<domain_name>.
The legal values for <ordering>
are:
- fixed
-
Records are returned in the order they are defined in the zone
file.
Note
The fixed option is only available if
BIND is configured with
--enable-fixed-rrset at compile time.
- random
Records are returned in a random order.
- cyclic
Records are returned in a cyclic round-robin order, rotating
by one record per query.
By default, records are returned in random order.
Note that if multiple rrset-order statements are
present in the configuration file (at both the
options and view levels), they
are not combined; instead, the more-specific
one (view) replaces the less-specific one
(options).
If multiple rules within a single rrset-order
statement match a given RRset, the first matching rule is applied.
Example:
rrset-order {
type A name "foo.isc.org" order random;
type AAAA name "foo.isc.org" order cyclic;
name "bar.isc.org" order fixed;
name "*.bar.isc.org" order random;
name "*.baz.isc.org" order cyclic;
};
With the above configuration, the following RRset ordering is used:
- lame-ttl
This is always set to 0. More information is available
in the security advisory for CVE-2021-25219.
- servfail-ttl
-
This sets the number of seconds to cache a
SERVFAIL response due to DNSSEC validation failure or
other general server failure. If set to
0, SERVFAIL caching is disabled.
The SERVFAIL cache is not consulted if a query has
the CD (Checking Disabled) bit set; this allows a
query that failed due to DNSSEC validation to be retried
without waiting for the SERVFAIL TTL to expire.
The maximum value is 30
seconds; any higher value is silently
reduced. The default is 1
second.
- max-ncache-ttl
To reduce network traffic and increase performance,
the server stores negative answers. max-ncache-ttl is
used to set a maximum retention time for these answers in
the server,
in seconds. The default
max-ncache-ttl is 10800 seconds (3 hours).
max-ncache-ttl cannot exceed
7 days and is
silently truncated to 7 days if set to a greater value.
- max-cache-ttl
This sets the maximum time for which the server
caches ordinary (positive) answers, in seconds.
The default is 604800 (one week).
A value of zero may cause all queries to return
SERVFAIL, because of lost caches of intermediate
RRsets (such as NS and glue AAAA/A records) in the
resolution process.
- min-roots
-
This sets the minimum number of root servers that
is required for a request for the root servers to be
accepted. The default
is 2.
Note
This is not implemented in BIND 9.
- sig-validity-interval
-
This specifies the number of days into the future that
DNSSEC signatures that are automatically generated as a
result of dynamic updates (the section called “Dynamic Update”) will expire. There
is an optional second field which specifies how
long before expiry that the signatures are
regenerated. If not specified, the signatures are
regenerated at 1/4 of base interval. The second
field is specified in days if the base interval is
greater than 7 days; otherwise it is specified in hours.
The default base interval is 30 days,
giving a re-signing interval of 7 1/2 days. The maximum
value is 10 years (3660 days).
The signature inception time is unconditionally
set to one hour before the current time, to allow
for a limited amount of clock skew.
The sig-validity-interval
should be at least several multiples of the SOA
expire interval, to allow for reasonable interaction
between the various timer and expiry dates.
- sig-signing-nodes
This specifies the maximum number of nodes to be
examined in each quantum, when signing a zone with
a new DNSKEY. The default is
100.
- sig-signing-signatures
This specifies a threshold number of signatures that
terminates processing a quantum, when signing
a zone with a new DNSKEY. The default is
10.
- sig-signing-type
-
This specifies a private RDATA type to be used when generating
signing-state records. The default is
65534.
This parameter may be removed
in a future version, once there is a standard type.
Signing-state records are used internally by
named to track the current state of
a zone-signing process, i.e., whether it is still active
or has been completed. The records can be inspected
using the command
rndc signing -list zone.
Once named has finished signing
a zone with a particular key, the signing-state
record associated with that key can be removed from
the zone by running
rndc signing -clear keyid/algorithm zone.
To clear all of the completed signing-state
records for a zone, use
rndc signing -clear all zone.
-
min-refresh-time, max-refresh-time, min-retry-time, max-retry-time
-
These options control the server's behavior on refreshing a
zone (querying for SOA changes) or retrying failed
transfers. Usually the SOA values for the zone are used,
up to a hard-coded maximum expiry of 24 weeks. However,
these values are set by the primary, giving secondary server
administrators little control over their contents.
These options allow the administrator to set a minimum and
maximum refresh and retry time in seconds per-zone,
per-view, or globally. These options are valid for
secondary and stub zones, and clamp the SOA refresh and
retry times to the specified values.
The following defaults apply:
min-refresh-time 300 seconds,
max-refresh-time 2419200 seconds
(4 weeks), min-retry-time 500 seconds,
and max-retry-time 1209600 seconds
(2 weeks).
- edns-udp-size
-
This sets the maximum advertised EDNS UDP buffer size, in
bytes, to control the size of packets received from
authoritative servers in response to recursive queries.
Valid values are 512 to 4096; values outside this range
are silently adjusted to the nearest value within
it. The default value is 1232.
The usual reason for setting
edns-udp-size to a non-default value
is to get UDP answers to pass through broken firewalls
that block fragmented packets and/or block UDP DNS
packets that are greater than 512 bytes.
When named first queries a remote
server, it advertises a UDP buffer size of 512, as
this has the greatest chance of success on the first try.
If the initial response times out, named
tries again with plain DNS; if that is successful,
it is taken as evidence that the server does not
support EDNS. After enough failures using EDNS and
successes using plain DNS, named
defaults to plain DNS for future communications
with that server. If that happens, named
periodically sends an EDNS query to see if the situation has
improved.
However, if the initial query is successful with
EDNS advertising a buffer size of 512, then
named advertises progressively
larger buffer sizes on successive queries, until
responses begin timing out or
edns-udp-size is reached.
The default buffer sizes used by named
are 512, 1232, 1432, and 4096, but never exceed
edns-udp-size. (The values 1232 and
1432 are chosen to allow for an IPv4-/IPv6-encapsulated
UDP message to be sent without fragmentation at the
minimum MTU sizes for Ethernet and IPv6 networks.)
- max-udp-size
-
This sets the maximum EDNS UDP message size that
named sends, in bytes.
Valid values are 512 to 4096; values outside this
range are silently adjusted to the nearest
value within it. The default value is 1232.
This value applies to responses sent by a server; to
set the advertised buffer size in queries, see
edns-udp-size.
The usual reason for setting
max-udp-size to a non-default
value is to allow UDP answers to pass through broken
firewalls that block fragmented packets and/or
block UDP packets that are greater than 512 bytes.
This is independent of the advertised receive
buffer (edns-udp-size).
Setting this to a low value encourages additional
TCP traffic to the name server.
- masterfile-format
-
This specifies
the file format of zone files (see
the section called “Additional File Formats”).
The default value is text, which is the
standard textual representation, except for secondary zones,
in which the default value is raw.
Files in formats other than text are
typically expected to be generated by the
named-compilezone tool, or dumped by
named.
Note that when a zone file in a format other than
text is loaded, named
may omit some of the checks which would be performed for a
file in text format. In particular,
check-names checks do not apply
for the raw format. This means
a zone file in the raw format
must be generated with the same check level as that
specified in the named configuration
file. Also, map format files are
loaded directly into memory via memory mapping, with only
minimal checking.
This statement sets the
masterfile-format for all zones,
but can be overridden on a per-zone or per-view basis
by including a masterfile-format
statement within the zone or
view block in the configuration
file.
- masterfile-style
-
This specifies the formatting of zone files during dump,
when the masterfile-format is
text. This option is ignored
with any other masterfile-format.
When set to relative,
records are printed in a multi-line format, with owner
names expressed relative to a shared origin. When set
to full, records are printed in
a single-line format with absolute owner names.
The full format is most suitable
when a zone file needs to be processed automatically
by a script. The relative format
is more human-readable, and is thus suitable when a
zone is to be edited by hand. The default is
relative.
-
max-recursion-depth
This sets the maximum number of levels of recursion
that are permitted at any one time while servicing
a recursive query. Resolving a name may require
looking up a name server address, which in turn
requires resolving another name, etc.; if the number
of recursions exceeds this value, the recursive
query is terminated and returns SERVFAIL. The
default is 7.
-
max-recursion-queries
This sets the maximum number of iterative queries that
may be sent while servicing a recursive query.
If more queries are sent, the recursive query
is terminated and returns SERVFAIL. The default is 100.
- notify-delay
-
This sets the delay, in seconds, between sending sets of NOTIFY
messages for a zone. The default is 5 seconds.
The overall rate at which NOTIFY messages are sent for all
zones is controlled by serial-query-rate.
- max-rsa-exponent-size
This sets the maximum RSA exponent size, in bits, that is
accepted when validating. Valid values are 35
to 4096 bits. The default, zero, is also accepted
and is equivalent to 4096.
- prefetch
-
When a query is received for cached data which
is to expire shortly, named can
refresh the data from the authoritative server
immediately, ensuring that the cache always has an
answer available.
prefetch specifies the
"trigger" TTL value at which prefetch of the current
query takes place; when a cache record with a
lower TTL value is encountered during query processing,
it is refreshed. Valid trigger TTL values are 1 to
10 seconds. Values larger than 10 seconds are silently
reduced to 10.
Setting a trigger TTL to zero causes
prefetch to be disabled.
The default trigger TTL is 2.
An optional second argument specifies the "eligibility"
TTL: the smallest original
TTL value that is accepted for a record to be
eligible for prefetching. The eligibility TTL must
be at least six seconds longer than the trigger TTL;
if not, named silently
adjusts it upward.
The default eligibility TTL is 9.
- v6-bias
When determining the next name server to try,
this indicates by how many milliseconds to prefer IPv6 name servers.
The default is 50 milliseconds.
Built-in Server Information Zones
The server provides some helpful diagnostic information
through a number of built-in zones under the
pseudo-top-level-domain bind in the
CHAOS class. These zones are part
of a
built-in view (see the section called “view Statement Grammar”) of
class
CHAOS, which is separate from the
default view of class IN. Most global
configuration options (allow-query,
etc.) apply to this view, but some are locally
overridden: notify,
recursion, and
allow-new-zones are
always set to no, and
rate-limit is set to allow
three responses per second.
To disable these zones, use the options
below or hide the built-in CHAOS
view by
defining an explicit view of class CHAOS
that matches all clients.
- version
-
This is the version the server should report
via a query of the name version.bind
with type TXT and class CHAOS.
The default is the real version number of this server.
Specifying version none
disables processing of the queries.
Setting version to any value
(including none) also
disables queries for authors.bind TXT CH.
- hostname
This is the hostname the server should report via a query of
the name hostname.bind
with type TXT and class CHAOS.
This defaults to the hostname of the machine hosting the
name server, as
found by the gethostname() function. The primary purpose of such queries
is to
identify which of a group of anycast servers is actually
answering the queries. Specifying hostname none;
disables processing of the queries.
- server-id
This is the ID the server should report when receiving a Name
Server Identifier (NSID) query, or a query of the name
ID.SERVER with type
TXT and class CHAOS.
The primary purpose of such queries is to
identify which of a group of anycast servers is actually
answering the queries. Specifying server-id none;
disables processing of the queries.
Specifying server-id hostname; causes named to
use the hostname as found by the gethostname() function.
The default server-id is none.
The named server has some built-in
empty zones, for SOA and NS records only.
These are for zones that should normally be answered locally
and which queries should not be sent to the Internet's root
servers. The official servers which cover these namespaces
return NXDOMAIN responses to these queries. In particular,
these cover the reverse namespaces for addresses from
RFC 1918, RFC 4193, RFC 5737, and RFC 6598. They also include the
reverse namespace for the IPv6 local address (locally assigned),
IPv6 link local addresses, the IPv6 loopback address, and the
IPv6 unknown address.
The server attempts to determine if a built-in zone
already exists or is active (covered by a forward-only
forwarding declaration) and does not create an empty
zone if either is true.
The current list of empty zones is:
- 10.IN-ADDR.ARPA
- 16.172.IN-ADDR.ARPA
- 17.172.IN-ADDR.ARPA
- 18.172.IN-ADDR.ARPA
- 19.172.IN-ADDR.ARPA
- 20.172.IN-ADDR.ARPA
- 21.172.IN-ADDR.ARPA
- 22.172.IN-ADDR.ARPA
- 23.172.IN-ADDR.ARPA
- 24.172.IN-ADDR.ARPA
- 25.172.IN-ADDR.ARPA
- 26.172.IN-ADDR.ARPA
- 27.172.IN-ADDR.ARPA
- 28.172.IN-ADDR.ARPA
- 29.172.IN-ADDR.ARPA
- 30.172.IN-ADDR.ARPA
- 31.172.IN-ADDR.ARPA
- 168.192.IN-ADDR.ARPA
- 64.100.IN-ADDR.ARPA
- 65.100.IN-ADDR.ARPA
- 66.100.IN-ADDR.ARPA
- 67.100.IN-ADDR.ARPA
- 68.100.IN-ADDR.ARPA
- 69.100.IN-ADDR.ARPA
- 70.100.IN-ADDR.ARPA
- 71.100.IN-ADDR.ARPA
- 72.100.IN-ADDR.ARPA
- 73.100.IN-ADDR.ARPA
- 74.100.IN-ADDR.ARPA
- 75.100.IN-ADDR.ARPA
- 76.100.IN-ADDR.ARPA
- 77.100.IN-ADDR.ARPA
- 78.100.IN-ADDR.ARPA
- 79.100.IN-ADDR.ARPA
- 80.100.IN-ADDR.ARPA
- 81.100.IN-ADDR.ARPA
- 82.100.IN-ADDR.ARPA
- 83.100.IN-ADDR.ARPA
- 84.100.IN-ADDR.ARPA
- 85.100.IN-ADDR.ARPA
- 86.100.IN-ADDR.ARPA
- 87.100.IN-ADDR.ARPA
- 88.100.IN-ADDR.ARPA
- 89.100.IN-ADDR.ARPA
- 90.100.IN-ADDR.ARPA
- 91.100.IN-ADDR.ARPA
- 92.100.IN-ADDR.ARPA
- 93.100.IN-ADDR.ARPA
- 94.100.IN-ADDR.ARPA
- 95.100.IN-ADDR.ARPA
- 96.100.IN-ADDR.ARPA
- 97.100.IN-ADDR.ARPA
- 98.100.IN-ADDR.ARPA
- 99.100.IN-ADDR.ARPA
- 100.100.IN-ADDR.ARPA
- 101.100.IN-ADDR.ARPA
- 102.100.IN-ADDR.ARPA
- 103.100.IN-ADDR.ARPA
- 104.100.IN-ADDR.ARPA
- 105.100.IN-ADDR.ARPA
- 106.100.IN-ADDR.ARPA
- 107.100.IN-ADDR.ARPA
- 108.100.IN-ADDR.ARPA
- 109.100.IN-ADDR.ARPA
- 110.100.IN-ADDR.ARPA
- 111.100.IN-ADDR.ARPA
- 112.100.IN-ADDR.ARPA
- 113.100.IN-ADDR.ARPA
- 114.100.IN-ADDR.ARPA
- 115.100.IN-ADDR.ARPA
- 116.100.IN-ADDR.ARPA
- 117.100.IN-ADDR.ARPA
- 118.100.IN-ADDR.ARPA
- 119.100.IN-ADDR.ARPA
- 120.100.IN-ADDR.ARPA
- 121.100.IN-ADDR.ARPA
- 122.100.IN-ADDR.ARPA
- 123.100.IN-ADDR.ARPA
- 124.100.IN-ADDR.ARPA
- 125.100.IN-ADDR.ARPA
- 126.100.IN-ADDR.ARPA
- 127.100.IN-ADDR.ARPA
- 0.IN-ADDR.ARPA
- 127.IN-ADDR.ARPA
- 254.169.IN-ADDR.ARPA
- 2.0.192.IN-ADDR.ARPA
- 100.51.198.IN-ADDR.ARPA
- 113.0.203.IN-ADDR.ARPA
- 255.255.255.255.IN-ADDR.ARPA
- 0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.IP6.ARPA
- 1.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.0.IP6.ARPA
- 8.B.D.0.1.0.0.2.IP6.ARPA
- D.F.IP6.ARPA
- 8.E.F.IP6.ARPA
- 9.E.F.IP6.ARPA
- A.E.F.IP6.ARPA
- B.E.F.IP6.ARPA
- EMPTY.AS112.ARPA
- HOME.ARPA
Empty zones can be set at the view level and only apply to
views of class IN. Disabled empty zones are only inherited
from options if there are no disabled empty zones specified
at the view level. To override the options list of disabled
zones, disable the root zone at the view level. For example:
disable-empty-zone ".";
If using the address ranges covered here,
reverse zones covering the addresses should already be in place.
In practice this appears to not be the case, with many queries
being made to the infrastructure servers for names in these
spaces. So many, in fact, that sacrificial servers had
to be deployed to channel the query load away from the
infrastructure servers.
Note
The real parent servers for these zones should disable all
empty zones under the parent zone they serve. For the real
root servers, this is all built-in empty zones. This
enables them to return referrals to deeper in the tree.
- empty-server
This specifies the server name that appears in the returned
SOA record for empty zones. If none is specified,
the zone's name is used.
- empty-contact
This specifies the contact name that appears in the returned
SOA record for empty zones. If none is specified,
"." is used.
- empty-zones-enable
This enables or disables all empty zones. By default, they
are enabled.
- disable-empty-zone
This disables individual empty zones. By default, none are
disabled. This option can be specified multiple times.
Additional Section Caching
The additional section cache, also called acache,
is an internal cache to improve the response performance of BIND 9.
When additional section caching is enabled, BIND 9
caches an internal shortcut to the additional section content for
each answer RR.
Note that acache is an internal caching
mechanism of BIND 9, and is not related to the DNS caching
server function.
Additional section caching does not change the
response content (except the RRsets ordering of the additional
section; see below), but can improve the response performance
significantly.
It is particularly effective when BIND 9 acts as an authoritative
server for a zone that has many delegations with many glue RRs.
To obtain the maximum performance improvement
from additional section caching, setting
additional-from-cache
to no is recommended, since the current
implementation of acache
does not shortcut additional section information from the
DNS cache data.
One obvious disadvantage of acache is
that it requires much more
memory for the internal cached data.
Thus, if the response performance does not matter and memory
consumption is more critical, the
acache mechanism can be
disabled by setting acache-enable to
no.
It is also possible to specify the upper limit of memory
consumption
for acache by using max-acache-size.
Additional section caching also has a minor effect on the
RRset ordering in the additional section.
Without acache,
cyclic order is effective for the additional
section as well as for the answer and authority sections.
However, additional section caching fixes the ordering when it
first caches an RRset for the additional section, and the same
ordering is kept in succeeding responses, regardless of the
setting of rrset-order.
The effect of this should be minor, however, since an
RRset in the additional section
typically only contains a small number of RRs (and in many cases
only a single RR), so the
ordering is not significant.
The following is a summary of options related to
acache.
- acache-enable
If yes, additional section caching is
enabled. The default value is no.
- acache-cleaning-interval
The server removes stale cache entries, based on an LRU-based
algorithm, every acache-cleaning-interval minutes.
The default is 60 minutes.
If set to 0, no periodic cleaning occurs.
- max-acache-size
This is the maximum amount of memory, in bytes, to use for the server's acache.
When the amount of data in the acache reaches this limit,
the server
cleans more aggressively so that the limit is not
exceeded.
In a server with multiple views, the limit applies
separately to the
acache of each view.
The default is 16M.
BIND 9 provides the ability to filter
out responses from external DNS servers containing
certain types of data in the answer section.
Specifically, it can reject address (A or AAAA) records if
the corresponding IPv4 or IPv6 addresses match the given
address_match_list of the
deny-answer-addresses option.
It can also reject CNAME or DNAME records if the "alias"
name (i.e., the CNAME alias or the substituted query name
due to DNAME) matches the
given namelist of the
deny-answer-aliases option, where
"match" means the alias name is a subdomain of one of
the name_list elements.
If the optional namelist is specified
with except-from, records whose query name
matches the list are accepted regardless of the filter
setting.
Likewise, if the alias name is a subdomain of the
corresponding zone, the deny-answer-aliases
filter does not apply;
for example, even if "example.com" is specified for
deny-answer-aliases,
www.example.com. CNAME xxx.example.com.
returned by an "example.com" server is accepted.
In the address_match_list of the
deny-answer-addresses option, only
ip_addr
and ip_prefix
are meaningful;
any key_id is silently ignored.
If a response message is rejected due to the filtering,
the entire message is discarded without being cached, and
a SERVFAIL error is returned to the client.
This filtering is intended to prevent "DNS rebinding attacks," in
which an attacker, in response to a query for a domain name the
attacker controls, returns an IP address within the user's own network or
an alias name within the user's own domain.
A naive web browser or script could then serve as an
unintended proxy, allowing the attacker
to get access to an internal node of the local network
that could not be externally accessed otherwise.
See the paper available at
https://dl.acm.org/doi/10.1145/1315245.1315298
for more details about these attacks.
For example, with a domain named "example.net" and
an internal network using an IPv4 prefix 192.0.2.0/24,
an administrator might specify the following rules:
deny-answer-addresses { 192.0.2.0/24; } except-from { "example.net"; };
deny-answer-aliases { "example.net"; };
If an external attacker let a web browser in the local
network look up an IPv4 address of "attacker.example.com",
the attacker's DNS server would return a response like this:
attacker.example.com. A 192.0.2.1
in the answer section.
Since the rdata of this record (the IPv4 address) matches
the specified prefix 192.0.2.0/24, this response would be
ignored.
On the other hand, if the browser looked up a legitimate
internal web server "www.example.net" and the
following response were returned to
the BIND 9 server:
www.example.net. A 192.0.2.2
it would be accepted, since the owner name "www.example.net"
matches the except-from element,
"example.net".
Note that this is not really an attack on the DNS per se.
In fact, there is nothing wrong with having an "external" name
mapped to an "internal" IP address or domain name
from the DNS point of view;
it might actually be provided for a legitimate purpose,
such as for debugging.
As long as the mapping is provided by the correct owner,
it either is not possible or does not make sense to detect
whether the intent of the mapping is legitimate
within the DNS.
The "rebinding" attack must primarily be protected at the
application that uses the DNS.
For a large site, however, it may be difficult to protect
all possible applications at once.
This filtering feature is provided only to help such an
operational environment;
turning it on is generally discouraged unless there is
no other choice and the attack is a
real threat to applications.
Care should be particularly taken if using this
option for addresses within 127.0.0.0/8.
These addresses are obviously "internal," but many
applications conventionally rely on a DNS mapping from
some name to such an address.
Filtering out DNS records containing this address
spuriously can break such applications.
Response Policy Zone (RPZ) Rewriting
BIND 9 includes a limited
mechanism to modify DNS responses for requests
analogous to email anti-spam DNS rejection lists.
Responses can be changed to deny the existence of domains (NXDOMAIN),
deny the existence of IP addresses for domains (NODATA),
or contain other IP addresses or data.
Response policy zones are named in the
response-policy option for the view or among the
global options if there is no response-policy option for the view.
Response policy zones are ordinary DNS zones containing RRsets
that can be queried normally if allowed.
It is usually best to restrict those queries with something like
allow-query { localhost; };.
Note that zones using masterfile-format map
cannot be used as policy zones.
A response-policy option can support
multiple policy zones. To maximize performance, a radix
tree is used to quickly identify response policy zones
containing triggers that match the current query. This
imposes an upper limit of 32 on the number of policy zones
in a single response-policy option; more
than that is a configuration error.
Rules encoded in response policy zones are processed after
those defined in Access Control Lists
(ACLs). All queries from clients which are not
permitted access to the resolver are answered with a
status code of REFUSED, regardless of configured RPZ rules.
Five policy triggers can be encoded in RPZ records.
- RPZ-CLIENT-IP
-
IP records are triggered by the IP address of the
DNS client.
Client IP address triggers are encoded in records that have
owner names that are subdomains of
rpz-client-ip, relativized to the
policy zone origin name,
and encode an address or address block.
IPv4 addresses are represented as
prefixlength.B4.B3.B2.B1.rpz-client-ip.
The IPv4 prefix length must be between 1 and 32.
All four bytes - B4, B3, B2, and B1 - must be present.
B4 is the decimal value of the least significant byte of the
IPv4 address as in IN-ADDR.ARPA.
IPv6 addresses are encoded in a format similar
to the standard IPv6 text representation,
prefixlength.W8.W7.W6.W5.W4.W3.W2.W1.rpz-client-ip.
Each of W8,...,W1 is a one- to four-digit hexadecimal number
representing 16 bits of the IPv6 address as in the standard
text representation of IPv6 addresses, but reversed as in
IP6.ARPA. (Note that this representation of IPv6
address is different from IP6.ARPA where each hex
digit occupies a label.)
All 8 words must be present except when one set of consecutive
zero words is replaced with .zz.,
analogous to double colons (::) in standard IPv6 text
encodings.
The IPv6 prefix length must be between 1 and 128.
- QNAME
QNAME policy records are triggered by query names of
requests and targets of CNAME records resolved to generate
the response.
The owner name of a QNAME policy record is
the query name relativized to the policy zone.
- RPZ-IP
IP triggers are IP addresses in an
A or AAAA record in the ANSWER section of a response.
They are encoded like client-IP triggers, except as
subdomains of rpz-ip.
- RPZ-NSDNAME
-
NSDNAME triggers match names of authoritative servers
for the query name, a parent of the query name, a CNAME for
the query name, or a parent of a CNAME.
They are encoded as subdomains of
rpz-nsdname, relativized
to the RPZ origin name.
NSIP triggers match IP addresses in A and
AAAA RRsets for domains that can be checked against NSDNAME
policy records. The
nsdname-enable phrase turns NSDNAME
triggers off or on for a single policy zone or for all
zones.
If authoritative nameservers for the query name are not
yet known, named recursively
looks up the authoritative servers for the query name
before applying an RPZ-NSDNAME rule,
which can cause a processing delay. To speed up
processing at the cost of precision, the
nsdname-wait-recurse option
can be used; when set to no,
RPZ-NSDNAME rules are only applied when authoritative
servers for the query name have already been looked up and
cached. If authoritative servers for the query name
are not in the cache, the RPZ-NSDNAME rule is
ignored, but the authoritative servers for the query name
are looked up in the background and the rule is
applied to subsequent queries. The default is
yes, meaning RPZ-NSDNAME
rules are always applied, even if authoritative
servers for the query name need to be looked up first.
- RPZ-NSIP
-
NSIP triggers match the IP addresses of authoritative
servers. They are enncoded like IP triggers, except as
subdomains of rpz-nsip.
NSDNAME and NSIP triggers are checked only for names with at
least min-ns-dots dots.
The default value of min-ns-dots is
1, to exclude top-level domains.
If a name server's IP address is not yet known,
named recursively looks up
the IP address before applying an RPZ-NSIP rule,
which can cause a processing delay. To speed up
processing at the cost of precision, the
nsip-wait-recurse option
can be used: when set to no,
RPZ-NSIP rules are only applied when a name
server's IP address has already been looked up and
cached. If a server's IP address is not in the
cache, the RPZ-NSIP rule is ignored,
but the address is looked up in the
background and the rule is applied
to subsequent queries. The default is
yes, meaning RPZ-NSIP
rules are always applied, even if an
address needs to be looked up first.
The query response is checked against all response policy zones,
so two or more policy records can be triggered by a response.
Because DNS responses are rewritten according to at most one
policy record, a single record encoding an action (other than
DISABLED actions) must be chosen.
Triggers, or the records that encode them, are chosen for
rewriting in the following order:
- Choose the triggered record in the zone that appears
first in the response-policy option.
- Prefer CLIENT-IP to QNAME to IP to NSDNAME to NSIP
triggers in a single zone.
- Among NSDNAME triggers, prefer the
trigger that matches the smallest name under the DNSSEC ordering.
- Among IP or NSIP triggers, prefer the trigger
with the longest prefix.
- Among triggers with the same prefix length,
prefer the IP or NSIP trigger that matches
the smallest IP address.
When the processing of a response is restarted to resolve
DNAME or CNAME records and a policy record set has
not been triggered,
all response policy zones are again consulted for the
DNAME or CNAME names and addresses.
RPZ record sets are any types of DNS record, except
DNAME or DNSSEC, that encode actions or responses to
individual queries.
Any of the policies can be used with any of the triggers.
For example, while the TCP-only policy is
commonly used with client-IP triggers,
it can be used with any type of trigger to force the use of
TCP for responses with owner names in a zone.
- PASSTHRU
The auto-acceptance policy is specified
by a CNAME whose target is rpz-passthru.
It causes the response to not be rewritten
and is most often used to "poke holes" in policies for
CIDR blocks.
- DROP
The auto-rejection policy is specified
by a CNAME whose target is rpz-drop.
It causes the response to be discarded.
Nothing is sent to the DNS client.
- TCP-Only
The "slip" policy is specified
by a CNAME whose target is rpz-tcp-only.
It changes UDP responses to short, truncated DNS responses
that require the DNS client to try again with TCP.
It is used to mitigate distributed DNS reflection attacks.
- NXDOMAIN
The "domain undefined" response is encoded
by a CNAME whose target is the root domain (.)
- NODATA
The empty set of resource records is specified by a
CNAME whose target is the wildcard top-level
domain (*.).
It rewrites the response to NODATA or ANCOUNT=0.
- Local Data
-
A set of ordinary DNS records can be used to answer queries.
Queries for record types not the set are answered with
NODATA.
A special form of local data is a CNAME whose target is a
wildcard such as *.example.com.
It is used as if an ordinary CNAME after the asterisk (*)
has been replaced with the query name.
This special form is useful for query logging in the
walled garden's authoritative DNS server.
All of the actions specified in all of the individual records
in a policy zone
can be overridden with a policy clause in the
response-policy option.
An organization using a policy zone provided by another
organization might use this mechanism to redirect domains
to its own walled garden.
- GIVEN
The placeholder policy says "do not override but
perform the action specified in the zone."
- DISABLED
The testing override policy causes policy zone records to do
nothing but log what they would have done if the
policy zone were not disabled.
The response to the DNS query is written (or not)
according to any triggered policy records that are not
disabled.
Disabled policy zones should appear first,
because they are often not logged
if a higher-precedence trigger is found first.
-
PASSTHRU, DROP, TCP-Only, NXDOMAIN, NODATA
each override the corresponding per-record policy.
- CNAME domain
causes all RPZ policy records to act as if they were
"cname domain" records.
By default, the actions encoded in a response policy zone
are applied only to queries that ask for recursion (RD=1).
That default can be changed for a single policy zone, or for
all response policy zones in a view,
with a recursive-only no clause.
This feature is useful for serving the same zone files
both inside and outside an RFC 1918 cloud and using RPZ to
delete answers that would otherwise contain RFC 1918 values
on the externally visible name server or view.
Also by default, RPZ actions are applied only to DNS requests
that either do not request DNSSEC metadata (DO=0) or when no
DNSSEC records are available for the requested name in the original
zone (not the response policy zone). This default can be
changed for all response policy zones in a view with a
break-dnssec yes clause. In that case, RPZ
actions are applied regardless of DNSSEC. The name of the
clause option reflects the fact that results rewritten by RPZ
actions cannot verify.
No DNS records are needed for a QNAME or Client-IP trigger;
the name or IP address itself is sufficient,
so in principle the query name need not be recursively resolved.
However, not resolving the requested
name can leak the fact that response policy rewriting is in use,
and that the name is listed in a policy zone, to operators of
servers for listed names. To prevent that information leak, by
default any recursion needed for a request is done before any
policy triggers are considered. Because listed domains often
have slow authoritative servers, this behavior can cost
significant time.
The qname-wait-recurse no option
overrides that default behavior when recursion cannot
change a non-error response.
The option does not affect QNAME or client-IP triggers
in policy zones listed
after other zones containing IP, NSIP, and NSDNAME triggers, because
those may depend on the A, AAAA, and NS records that would be
found during recursive resolution. It also does not affect
DNSSEC requests (DO=1) unless break-dnssec yes
is in use, because the response would depend on whether
RRSIG records were found during resolution.
Using this option can cause error responses such as SERVFAIL to
appear to be rewritten, since no recursion is being done to
discover problems at the authoritative server.
The TTL of a record modified by RPZ policies is set from the
TTL of the relevant record in the policy zone. It is then limited
to a maximum value.
The max-policy-ttl clause changes the
maximum number of seconds from its default of 5.
For example, an administrator might use this option statement:
response-policy { zone "badlist"; };
and this zone statement:
zone "badlist" {type master; file "master/badlist"; allow-query {none;}; };
with this zone file:
$TTL 1H
@ SOA LOCALHOST. named-mgr.example.com (1 1h 15m 30d 2h)
NS LOCALHOST.
; QNAME policy records. There are no periods (.) after the owner names.
nxdomain.domain.com CNAME . ; NXDOMAIN policy
*.nxdomain.domain.com CNAME . ; NXDOMAIN policy
nodata.domain.com CNAME *. ; NODATA policy
*.nodata.domain.com CNAME *. ; NODATA policy
bad.domain.com A 10.0.0.1 ; redirect to a walled garden
AAAA 2001:2::1
bzone.domain.com CNAME garden.example.com.
; do not rewrite (PASSTHRU) OK.DOMAIN.COM
ok.domain.com CNAME rpz-passthru.
; redirect x.bzone.domain.com to x.bzone.domain.com.garden.example.com
*.bzone.domain.com CNAME *.garden.example.com.
; IP policy records that rewrite all responses containing A records in 127/8
; except 127.0.0.1
8.0.0.0.127.rpz-ip CNAME .
32.1.0.0.127.rpz-ip CNAME rpz-passthru.
; NSDNAME and NSIP policy records
ns.domain.com.rpz-nsdname CNAME .
48.zz.2.2001.rpz-nsip CNAME .
; auto-reject and auto-accept some DNS clients
112.zz.2001.rpz-client-ip CNAME rpz-drop.
8.0.0.0.127.rpz-client-ip CNAME rpz-drop.
; force some DNS clients and responses in the example.com zone to TCP
16.0.0.1.10.rpz-client-ip CNAME rpz-tcp-only.
example.com CNAME rpz-tcp-only.
*.example.com CNAME rpz-tcp-only.
RPZ can affect server performance.
Each configured response policy zone requires the server to
perform one to four additional database lookups before a
query can be answered.
For example, a DNS server with four policy zones, each with all
four kinds of response triggers (QNAME, IP, NSIP, and
NSDNAME), requires a total of 17 times as many database
lookups as a similar DNS server with no response policy zones.
A BIND 9 server with adequate memory and one
response policy zone with QNAME and IP triggers might achieve a
maximum queries-per-second (QPS) rate about 20% lower.
A server with four response policy zones with QNAME and IP
triggers might have a maximum QPS rate about 50% lower.
Responses rewritten by RPZ are counted in the
RPZRewrites statistics.
The log clause can be used to optionally
turn off rewrite logging for a particular response policy
zone. By default, all rewrites are logged.
Excessive, almost identical UDP responses
can be controlled by configuring a
rate-limit clause in an
options or view statement.
This mechanism keeps authoritative BIND 9 from being used
to amplify reflection denial of service (DoS) attacks.
Short, truncated (TC=1) responses can be sent to provide
rate-limited responses to legitimate clients within
a range of forged, attacked IP addresses.
Legitimate clients react to dropped or truncated responses
by retrying with UDP or with TCP, respectively.
This mechanism is intended for authoritative DNS servers.
It can be used on recursive servers, but can slow
applications such as SMTP servers (mail receivers) and
HTTP clients (web browsers) that repeatedly request the
same domains.
When possible, closing "open" recursive servers is better.
Response rate limiting uses a "credit" or "token bucket" scheme.
Each combination of identical response and client
has a conceptual "account" that earns a specified number
of credits every second.
A prospective response debits its account by one.
Responses are dropped or truncated
while the account is negative.
Responses are tracked within a rolling window of time
which defaults to 15 seconds, but which can be configured with
the window option to any value from
1 to 3600 seconds (1 hour).
The account cannot become more positive than
the per-second limit
or more negative than window
times the per-second limit.
When the specified number of credits for a class of
responses is set to 0, those responses are not rate-limited.
The notions of "identical response" and "DNS client"
for rate limiting are not simplistic.
All responses to an address block are counted as if to a
single client.
The prefix lengths of address blocks are
specified with ipv4-prefix-length (default 24)
and ipv6-prefix-length (default 56).
All non-empty responses for a valid domain name (qname)
and record type (qtype) are identical and have a limit specified
with responses-per-second
(default 0 or no limit).
All empty (NODATA) responses for a valid domain,
regardless of query type, are identical.
Responses in the NODATA class are limited by
nodata-per-second
(default responses-per-second).
Requests for any and all undefined subdomains of a given
valid domain result in NXDOMAIN errors, and are identical
regardless of query type.
They are limited by nxdomains-per-second
(default responses-per-second).
This controls some attacks using random names, but
can be relaxed or turned off (set to 0)
on servers that expect many legitimate
NXDOMAIN responses, such as from anti-spam rejection lists.
Referrals or delegations to the server of a given
domain are identical and are limited by
referrals-per-second
(default responses-per-second).
Responses generated from local wildcards are counted and limited
as if they were for the parent domain name.
This controls flooding using random.wild.example.com.
All requests that result in DNS errors other
than NXDOMAIN, such as SERVFAIL and FORMERR, are identical
regardless of requested name (qname) or record type (qtype).
This controls attacks using invalid requests or distant,
broken authoritative servers.
By default the limit on errors is the same as the
responses-per-second value,
but it can be set separately with
errors-per-second.
Many attacks using DNS involve UDP requests with forged source
addresses.
Rate limiting prevents the use of BIND 9 to flood a network
with responses to requests with forged source addresses,
but could let a third party block responses to legitimate requests.
There is a mechanism that can answer some legitimate
requests from a client whose address is being forged in a flood.
Setting slip to 2 (its default) causes every
other UDP request to be answered with a small truncated (TC=1)
response.
The small size and reduced frequency, and resulting lack of
amplification, of "slipped" responses make them unattractive
for reflection DoS attacks.
slip must be between 0 and 10.
A value of 0 does not "slip";
no truncated responses are sent due to rate limiting.
Rather, all responses are dropped.
A value of 1 causes every response to slip;
values between 2 and 10 cause every nth response to slip.
Some error responses, including REFUSED and SERVFAIL,
cannot be replaced with truncated responses and are instead
leaked at the slip rate.
(Note: dropped responses from an authoritative server may
reduce the difficulty of a third party successfully forging
a response to a recursive resolver. The best security
against forged responses is for authoritative operators
to sign their zones using DNSSEC and for resolver operators
to validate the responses. When this is not an option,
operators who are more concerned with response integrity
than with flood mitigation may consider setting
slip to 1, causing all rate-limited
responses to be truncated rather than dropped. This reduces
the effectiveness of rate-limiting against reflection attacks.)
When the approximate query-per-second rate exceeds
the qps-scale value,
the responses-per-second,
errors-per-second,
nxdomains-per-second, and
all-per-second values are reduced by the
ratio of the current rate to the qps-scale value.
This feature can tighten defenses during attacks.
For example, with
qps-scale 250; responses-per-second 20; and
a total query rate of 1000 queries/second for all queries from
all DNS clients including via TCP,
then the effective responses/second limit changes to
(250/1000)*20, or 5.
Responses sent via TCP are not limited
but are counted to compute the query-per-second rate.
Communities of DNS clients can be given their own parameters or no
rate limiting by putting
rate-limit statements in view
statements instead of in the global option
statement.
A rate-limit statement in a view replaces,
rather than supplements, a rate-limit
statement among the main options.
DNS clients within a view can be exempted from rate limits
with the exempt-clients clause.
UDP responses of all kinds can be limited with the
all-per-second phrase. This rate
limiting is unlike the rate limiting provided by
responses-per-second,
errors-per-second, and
nxdomains-per-second on a DNS server,
which are often invisible to the victim of a DNS
reflection attack. Unless the forged requests of the
attack are the same as the legitimate requests of the
victim, the victim's requests are not affected. Responses
affected by an all-per-second limit
are always dropped; the slip value
has no effect. An all-per-second
limit should be at least 4 times as large as the other
limits, because single DNS clients often send bursts
of legitimate requests. For example, the receipt of a
single mail message can prompt requests from an SMTP
server for NS, PTR, A, and AAAA records as the incoming
SMTP/TCP/IP connection is considered. The SMTP server
can need additional NS, A, AAAA, MX, TXT, and SPF records
as it considers the SMTP Mail From
command. Web browsers often repeatedly resolve the
same names that are duplicated in HTML <IMG> tags
in a page. all-per-second is similar
to the rate limiting offered by firewalls but is often
inferior. Attacks that justify ignoring the contents
of DNS responses are likely to be attacks on the DNS
server itself. They usually should be discarded before
the DNS server spends resources make TCP connections
or parsing DNS requests, but that rate limiting must
be done before the DNS server sees the requests.
The maximum size of the table used to track requests and
rate-limit responses is set with max-table-size.
Each entry in the table is between 40 and 80 bytes.
The table needs approximately as many entries as the number
of requests received per second.
The default is 20,000.
To reduce the cold start of growing the table,
min-table-size (default 500)
can set the minimum table size.
Enable rate-limit category logging to monitor
expansions of the table and inform
choices for the initial and maximum table size.
Use log-only yes to test rate-limiting parameters
without actually dropping any requests.
Responses dropped by rate limits are included in the
RateDropped and QryDropped
statistics.
Responses that truncated by rate limits are included in
RateSlipped and RespTruncated.
named supports NXDOMAIN redirection via two methods:
With either method, when named gets an NXDOMAIN response
it examines a separate namespace to see if the NXDOMAIN
response should be replaced with an alternative response.
With a redirect zone (zone "." { type redirect; };), the
data used to replace the NXDOMAIN is held in a single
zone which is not part of the normal namespace. All the
redirect information is contained in the zone; there are
no delegations.
With a redirect namespace (option { nxdomain-redirect
<suffix> };), the data used to replace the
NXDOMAIN is part of the normal namespace and is looked up by
appending the specified suffix to the original query name.
This roughly doubles the cache required to process NXDOMAIN
responses, as both the original NXDOMAIN response and
the replacement data (or a NXDOMAIN indicating that there
is no replacement) must be stored.
If both a redirect zone and a redirect namespace are configured,
the redirect zone is tried first.
server Statement Definition and
Usage
The server statement defines
characteristics
to be associated with a remote name server. If a prefix length is
specified, then a range of servers is covered. Only the most
specific
server clause applies, regardless of the order in
named.conf.
The server statement can occur at
the top level of the
configuration file or inside a view
statement.
If a view statement contains
one or more server statements, only
those
apply to the view and any top-level ones are ignored.
If a view contains no server
statements,
any top-level server statements are
used as
defaults.
If a remote server is giving out bad data,
marking it as bogus prevents further queries to it. The
default
value of bogus is no.
The provide-ixfr clause determines
whether
the local server, acting as primary, responds with an
incremental
zone transfer when the given remote server, a secondary, requests it.
If set to yes, incremental transfer
is provided
whenever possible. If set to no,
all transfers
to the remote server are non-incremental. If not set, the
value
of the provide-ixfr option in the
view or
global options block is used as a default.
The request-ixfr clause determines
whether
the local server, acting as a secondary, requests incremental zone
transfers from the given remote server, a primary. If not set, the
value of the request-ixfr option in
the view or global options block is used as a default. It may
also be set in the zone block; if set there, it
overrides the global or view setting for that zone.
IXFR requests to servers that do not support IXFR
automatically
fall back to AXFR. Therefore, there is no need to manually list
which servers support IXFR and which ones do not; the global
default
of yes should always work.
The purpose of the provide-ixfr and
request-ixfr clauses is
to make it possible to disable the use of IXFR even when both
primary
and secondary claim to support it: for example, if one of the servers
is buggy and crashes or corrupts data when IXFR is used.
The request-expire clause determines
whether the local server, when acting as a secondary,
requests the EDNS EXPIRE value. The EDNS EXPIRE value
indicates the remaining time before the zone data
expires and needs to be refreshed. This is used
when a secondary server transfers a zone from another
secondary server; when transferring from the primary, the
expiration timer is set from the EXPIRE field of the SOA
record instead.
The default is yes.
The edns clause determines whether
the local server attempts to use EDNS when communicating
with the remote server. The default is yes.
The edns-udp-size option sets the
EDNS UDP size that is advertised by named
when querying the remote server. Valid values are 512
to 4096 bytes; values outside this range are silently
adjusted to the nearest value within it. This option
is useful when advertising a different value
to this server than the value advertised globally:
for example, when there is a firewall at the remote
site that is blocking large replies. Note: currently,
this sets a single UDP size for all packets sent to the
server; named does not deviate from
this value. This differs from the behavior of
edns-udp-size in options
or view statements, where it specifies
a maximum value. The server statement
behavior may be brought into conformance with the
options/view behavior in future releases.
The edns-version option sets the
maximum EDNS VERSION that is sent to the server(s)
by the resolver. The actual EDNS version sent is still
subject to normal EDNS version-negotiation rules (see
RFC 6891), the maximum EDNS version supported by the
server, and any other heuristics that indicate that a
lower version should be sent. This option is intended
to be used when a remote server reacts badly to a given
EDNS version or higher; it should be set to the highest
version the remote server is known to support. Valid
values are 0 to 255; higher values are silently
adjusted. This option is not needed until higher
EDNS versions than 0 are in use.
The max-udp-size option sets the
maximum EDNS UDP message size named
sends. Valid values are 512 to 4096 bytes; values
outside this range are silently adjusted. This
option is useful when there is a firewall
that is blocking large replies from named.
The tcp-only option sets the transport
protocol to TCP. The default is to use the UDP transport
and to fallback on TCP only when a truncated response
is received.
The server supports two zone transfer methods. The first, one-answer,
uses one DNS message per resource record transferred. many-answers packs
as many resource records as possible into a single message, which is
more efficient. It is possible to specify which method
to use for a server via the transfer-format option;
If not set there, the transfer-format
specified
by the options statement is
used.
transfers
is used to limit the number of concurrent inbound zone
transfers from the specified server. If no
transfers clause is specified, the
limit is set according to the
transfers-per-ns option.
The keys clause identifies a
key_id defined by the key statement,
to be used for transaction security (TSIG, the section called “TSIG”)
when talking to the remote server.
When a request is sent to the remote server, a request signature
is generated using the key specified here and appended to the
message. A request originating from the remote server is not
required
to be signed by this key.
Only a single key per server is currently supported.
The transfer-source and
transfer-source-v6 clauses specify
the IPv4 and IPv6 source
address, respectively, to be used for zone transfer with the remote server.
For an IPv4 remote server, only transfer-source can
be specified.
Similarly, for an IPv6 remote server, only
transfer-source-v6 can be
specified.
For more details, see the description of
transfer-source and
transfer-source-v6 in
the section called “Zone Transfers”.
The notify-source and
notify-source-v6 clauses specify the
IPv4 and IPv6 source address, respectively, to be used for notify
messages sent to remote servers. For an
IPv4 remote server, only notify-source
can be specified. Similarly, for an IPv6 remote server,
only notify-source-v6 can be specified.
The query-source and
query-source-v6 clauses specify the
IPv4 and IPv6 source address, respectively, to be used for queries
sent to remote servers. For an IPv4
remote server, only query-source can
be specified. Similarly, for an IPv6 remote server,
only query-source-v6 can be specified.
The request-nsid clause determines
whether the local server adds an NSID EDNS option
to requests sent to the server. This overrides
request-nsid set at the view or
option level.
The send-cookie clause determines
whether the local server adds a COOKIE EDNS option
to requests sent to the server. This overrides
send-cookie set at the view or
option level. The named server may
determine that COOKIE is not supported by the remote server
and not add a COOKIE EDNS option to requests.
managed-keys Statement Definition
and Usage
The managed-keys statement, like
trusted-keys, defines DNSSEC
security roots. The difference is that
managed-keys can be kept up-to-date
automatically, without intervention from the resolver
operator.
Suppose, for example, that a zone's key-signing
key was compromised, and the zone owner had to revoke and
replace the key. A resolver which had the old key in a
trusted-keys statement would be
unable to validate this zone; it would
reply with a SERVFAIL response code. This would
continue until the resolver operator updated the
trusted-keys statement with the new key.
If, however, the zone were listed in a
managed-keys statement instead, the
zone owner could add a "stand-by" key to the zone in advance.
named would store the stand-by key, and
when the original key was revoked, named
would be able to transition smoothly to the new key. It would
also recognize that the old key had been revoked and cease
using that key to validate answers, minimizing the damage that
the compromised key could do.
A managed-keys statement contains a list of
the keys to be managed, along with information about how the
keys are to be initialized for the first time. The only
initialization method currently supported is
initial-key.
This means the managed-keys statement must
contain a copy of the initializing key. (Future releases may
allow keys to be initialized by other methods, eliminating this
requirement.)
Consequently, a managed-keys statement
appears similar to a trusted-keys statement, differing
by the presence of the second field, which contains the keyword
initial-key. The difference is, whereas the
keys listed in a trusted-keys continue to be
trusted until they are removed from
named.conf, an initializing key listed
in a managed-keys statement is only trusted
once: for as long as it takes to load the
managed-key database and start the RFC 5011 key-maintenance
process.
The first time named runs with a managed key
configured in named.conf, it fetches the
DNSKEY RRset directly from the zone apex, and validates it
using the key specified in the managed-keys
statement. If the DNSKEY RRset is validly signed, then it is
used as the basis for a new managed-keys database.
From that point on, whenever named runs, it
sees the managed-keys statement, checks to
make sure RFC 5011 key maintenance has already been initialized
for the specified domain, and if so, simply moves on. The
key specified in the managed-keys
statement is not used to validate answers; it is
superseded by the key or keys stored in the managed-keys database.
The next time named runs after a name
has been removed from the
managed-keys statement, the corresponding
zone is removed from the managed-keys database,
and RFC 5011 key maintenance is no longer used for that
domain.
In the current implementation, the managed-keys database
is stored as a master-format zone file.
On servers which do not use views, this file is named
managed-keys.bind. When views are in
use, there is a separate managed-keys database for each
view; the filename is the view name (or, if a view name
contains characters which would make it illegal as a filename,
a hash of the view name), followed by
the suffix .mkeys.
When the key database is changed, the zone is updated.
As with any other dynamic zone, changes are written
into a journal file, e.g.,
managed-keys.bind.jnl or
internal.mkeys.jnl.
Changes are committed to the zone file as soon as
possible afterward, usually within 30
seconds. Whenever named is using
automatic key maintenance, the zone file and journal file
can be expected to exist in the working directory.
(For this reason, among others, the working directory
should be always be writable by named.)
If the dnssec-validation option is
set to auto, named
automatically initializes a managed key for the
root zone. The key that is used to initialize the
key-maintenance process is stored in bind.keys;
the location of this file can be overridden with the
bindkeys-file option. As a fallback
in the event no bind.keys can be
found, the initializing key is also compiled directly
into named.
zone Statement Definition and Usage
The type keyword is required
for the zone configuration unless
it is an in-view configuration. Its
acceptable values are: master,
slave, hint,
stub, static-stub,
forward, redirect,
or delegation-only.
Note
Later versions of BIND added type primary
and type secondary as synonyms for
type master and type slave,
as those terms are in more common use now. BIND 9.11's
configuration syntax predates this change.
The zone's name may optionally be followed by a class. If
a class is not specified, class IN (for Internet),
is assumed. This is correct for the vast majority of cases.
The hesiod class is
named for an information service from MIT's Project Athena. It was
used to share information about various systems databases, such
as users, groups, printers, and so on. The keyword
HS is
a synonym for hesiod.
Another MIT development is Chaosnet, a LAN protocol created
in the mid-1970s. Zone data for it can be specified with the CHAOS class.
- allow-notify
See the description of
allow-notify in the section called “Access Control”.
- allow-query
See the description of
allow-query in the section called “Access Control”.
- allow-query-on
See the description of
allow-query-on in the section called “Access Control”.
- allow-transfer
See the description of allow-transfer
in the section called “Access Control”.
- allow-update
See the description of allow-update
in the section called “Access Control”.
- update-policy
This specifies a "Simple Secure Update" policy. See
the section called “Dynamic Update Policies”.
- allow-update-forwarding
See the description of allow-update-forwarding
in the section called “Access Control”.
- also-notify
This option is only meaningful if notify
is
active for this zone. The set of machines that
receive a
DNS NOTIFY message
for this zone is made up of all the listed name servers
(other than
the primary) for the zone, plus any IP addresses
specified
with also-notify. A port
may be specified
with each also-notify
address to send the notify
messages to a port other than the default of 53.
A TSIG key may also be specified to cause the
NOTIFY to be signed by the
given key.
also-notify is not
meaningful for stub zones.
The default is the empty list.
- check-names
This option is used to restrict the character set and
syntax of
certain domain names in zone files and/or DNS responses
received from the
network. The default varies according to zone type.
For primary zones the default is fail;
for secondary zones the default is warn.
It is not implemented for hint zones.
- check-mx
See the description of
check-mx in the section called “Boolean Options”.
- check-spf
See the description of
check-spf in the section called “Boolean Options”.
- check-wildcard
See the description of
check-wildcard in the section called “Boolean Options”.
- check-integrity
See the description of
check-integrity in the section called “Boolean Options”.
- check-sibling
See the description of
check-sibling in the section called “Boolean Options”.
- zero-no-soa-ttl
See the description of
zero-no-soa-ttl in the section called “Boolean Options”.
- update-check-ksk
See the description of
update-check-ksk in the section called “Boolean Options”.
- dnssec-loadkeys-interval
See the description of
dnssec-loadkeys-interval in the section called “options Statement Definition and
Usage”.
- dnssec-update-mode
See the description of
dnssec-update-mode in the section called “options Statement Definition and
Usage”.
- dnssec-dnskey-kskonly
See the description of
dnssec-dnskey-kskonly in the section called “Boolean Options”.
- try-tcp-refresh
See the description of
try-tcp-refresh in the section called “Boolean Options”.
- database
-
This specifies the type of database to be used to store the
zone data. The string following the database keyword
is interpreted as a list of whitespace-delimited words.
The first word
identifies the database type, and any subsequent words are
passed
as arguments to the database to be interpreted in a way
specific
to the database type.
The default is "rbt", BIND 9's
native in-memory
red-black tree database. This database does not take
arguments.
Other values are possible if additional database drivers
have been linked into the server. Some sample drivers are
included
with the distribution but none are linked in by default.
- dialup
See the description of
dialup in the section called “Boolean Options”.
- delegation-only
-
This flag only applies to forward, hint, and stub
zones. If set to yes,
then the zone is treated as if it is
also a delegation-only type zone.
See caveats in root-delegation-only.
- file
This sets the zone's filename. In master,
hint, and redirect
zones which do not have masters
defined, zone data is loaded from this file. In
slave, stub, and
redirect zones which do have
masters defined, zone data is
retrieved from another server and saved in this file.
This option is not applicable to other zone types.
- forward
This option is only meaningful if the zone has a forwarders
list. The only value causes
the lookup to fail
after trying the forwarders and getting no answer, while first
allows a normal lookup to be tried.
- forwarders
This is used to override the list of global forwarders.
If it is not specified in a zone of type forward,
no forwarding is done for the zone and the global options are
not used.
- ixfr-base
This was used in BIND 8 to
specify the name
of the transaction log (journal) file for dynamic update
and IXFR.
BIND 9 ignores the option
and constructs the name of the journal
file by appending ".jnl"
to the name of the
zone file.
- ixfr-tmp-file
This was an undocumented option in BIND 8.
It is ignored in BIND 9.
- journal
This allows the default journal's filename to be overridden.
The default is the zone's filename with ".jnl" appended.
This is applicable to primary (master)
and secondary (slave) zones.
- max-journal-size
See the description of
max-journal-size in the section called “Server Resource Limits”.
- max-records
See the description of
max-records in the section called “Server Resource Limits”.
- max-transfer-time-in
See the description of
max-transfer-time-in in the section called “Zone Transfers”.
- max-transfer-idle-in
See the description of
max-transfer-idle-in in the section called “Zone Transfers”.
- max-transfer-time-out
See the description of
max-transfer-time-out in the section called “Zone Transfers”.
- max-transfer-idle-out
See the description of
max-transfer-idle-out in the section called “Zone Transfers”.
- notify
See the description of
notify in the section called “Boolean Options”.
- notify-delay
See the description of
notify-delay in the section called “Tuning”.
- notify-to-soa
See the description of
notify-to-soa in
the section called “Boolean Options”.
- pubkey
In BIND 8, this option was
intended to specify
a public zone key for verification of signatures in DNSSEC-signed
zones when they were loaded from disk. BIND 9 does not verify signatures
on load and ignores the option.
- zone-statistics
See the description of
zone-statistics in
the section called “options Statement Definition and
Usage”.
- server-addresses
-
This option is only meaningful for static-stub zones.
This is a list of IP addresses to which queries
should be sent in recursive resolution for the
zone.
A non-empty list for this option internally
configures the apex NS RR with associated glue A or
AAAA RRs.
For example, if "example.com" is configured as a
static-stub zone with 192.0.2.1 and 2001:db8::1234
in a server-addresses option,
the following RRs are internally configured:
example.com. NS example.com.
example.com. A 192.0.2.1
example.com. AAAA 2001:db8::1234
These records are used internally to resolve
names under the static-stub zone.
For instance, if the server receives a query for
"www.example.com" with the RD bit on, the server
initiates recursive resolution and sends
queries to 192.0.2.1 and/or 2001:db8::1234.
- server-names
-
This option is only meaningful for static-stub zones.
This is a list of domain names of name servers that
act as authoritative servers of the static-stub
zone.
These names are resolved to IP addresses when
named needs to send queries to
these servers.
For this supplemental resolution to be successful,
these names must not be a subdomain of the origin
name of the static-stub zone.
That is, when "example.net" is the origin of a
static-stub zone, "ns.example" and
"master.example.com" can be specified in the
server-names option, but
"ns.example.net" cannot; it is rejected by
the configuration parser.
A non-empty list for this option internally
configures the apex NS RR with the specified names.
For example, if "example.com" is configured as a
static-stub zone with "ns1.example.net" and
"ns2.example.net"
in a server-names option,
the following RRs are internally configured:
example.com. NS ns1.example.net.
example.com. NS ns2.example.net.
These records are used internally to resolve
names under the static-stub zone.
For instance, if the server receives a query for
"www.example.com" with the RD bit on, the server
initiates recursive resolution,
resolves "ns1.example.net" and/or
"ns2.example.net" to IP addresses, and then sends
queries to one or more of these addresses.
- sig-validity-interval
See the description of
sig-validity-interval in the section called “Tuning”.
- sig-signing-nodes
See the description of
sig-signing-nodes in the section called “Tuning”.
- sig-signing-signatures
See the description of
sig-signing-signatures in the section called “Tuning”.
- sig-signing-type
See the description of
sig-signing-type in the section called “Tuning”.
- transfer-source
See the description of
transfer-source in the section called “Zone Transfers”.
- transfer-source-v6
See the description of
transfer-source-v6 in the section called “Zone Transfers”.
- alt-transfer-source
See the description of
alt-transfer-source in the section called “Zone Transfers”.
- alt-transfer-source-v6
See the description of
alt-transfer-source-v6 in the section called “Zone Transfers”.
- use-alt-transfer-source
See the description of
use-alt-transfer-source in the section called “Zone Transfers”.
- notify-source
See the description of
notify-source in the section called “Zone Transfers”.
- notify-source-v6
See the description of
notify-source-v6 in the section called “Zone Transfers”.
-
min-refresh-time, max-refresh-time, min-retry-time, max-retry-time
See the descriptions in the section called “Tuning”.
- ixfr-from-differences
See the description of
ixfr-from-differences in the section called “Boolean Options”.
(Note that the ixfr-from-differences
choices of master and
slave are not
available at the zone level.)
- key-directory
See the description of
key-directory in the section called “options Statement Definition and
Usage”.
- auto-dnssec
See the description of
auto-dnssec in
the section called “options Statement Definition and
Usage”.
- serial-update-method
See the description of
serial-update-method in
the section called “options Statement Definition and
Usage”.
- inline-signing
If yes, this enables
"bump in the wire" signing of a zone, where an
unsigned zone is transferred in or loaded from
disk and a signed version of the zone is served,
with, possibly, a different serial number. This
behavior is disabled by default.
- multi-master
See the description of multi-master in
the section called “Boolean Options”.
- masterfile-format
See the description of masterfile-format
in the section called “Tuning”.
- max-zone-ttl
See the description of max-zone-ttl
in the section called “options Statement Definition and
Usage”.
- dnssec-secure-to-insecure
See the description of
dnssec-secure-to-insecure in the section called “Boolean Options”.
BIND 9 supports two
methods of granting clients the right to perform
dynamic updates to a zone, configured by the
allow-update and
update-policy options.
The allow-update clause is a simple
access control list. Any client that matches
the ACL is granted permission to update any record
in the zone.
The update-policy clause
allows more fine-grained control over which updates are
allowed. It specifies a set of rules, in which each rule
either grants or denies permission for one or more
names in the zone to be updated by one or more
identities. Identity is determined by the key that
signed the update request, using either TSIG or SIG(0).
In most cases, update-policy rules
only apply to key-based identities. There is no way
to specify update permissions based on client source
address.
update-policy rules are only meaningful
for primary zones (type master), and are
not allowed in any other zone type.
It is a configuration error to specify both
allow-update and
update-policy at the same time.
A pre-defined update-policy rule can be
switched on with the command
update-policy local;.
named automatically generates a TSIG session
key when starting and stores it in a file; this key can then
be used by local clients to update the zone while
named is running.
By default, the session key is stored in the file
/var/run/named/session.key, the key name
is "local-ddns", and the key algorithm is HMAC-SHA256.
These values are configurable with the
session-keyfile,
session-keyname, and
session-keyalg options, respectively.
A client running on the local system, if run with appropriate
permissions, may read the session key from the key file and
use it to sign update requests. The zone's update
policy is set to allow that key to change any record
within the zone. Assuming the key name is "local-ddns",
this policy is equivalent to:
update-policy { grant local-ddns zonesub any; };
with the additional restriction that only clients
connecting from the local system are permitted to send
updates.
Note that only one session key is generated by
named; all zones configured to use
update-policy local accept the same key.
The command nsupdate -l implements this
feature, sending requests to localhost and signing them using
the key retrieved from the session key file.
Other rule definitions look like this:
( grant | deny ) identity ruletype [ name ] [ types ]
Each rule grants or denies privileges. Rules are checked
in the order in which they are specified in the
update-policy statement. Once a message
has successfully matched a rule, the operation is immediately
granted or denied, and no further rules are examined. There
are 13 types of rules; the rule type is specified by the
ruletype field, and the interpretation
of other fields varies depending on the rule type.
In general, a rule is matched when the
key that signed an update request matches the
identity field, the name of the record
to be updated matches the name field
(in the manner specified by the ruletype
field), and the type of the record to be updated matches the
types field. Details for each rule type
are described below.
The identity field must be set to
a fully qualified domain name. In most cases, this
represents the name of the TSIG or SIG(0) key that must be
used to sign the update request. If the specified name is a
wildcard, it is subject to DNS wildcard expansion, and the
rule may apply to multiple identities. When a TKEY exchange
has been used to create a shared secret, the identity of
the key used to authenticate the TKEY exchange is
used as the identity of the shared secret. Some rule types
use identities matching the client's Kerberos principal
(e.g, "host/machine@REALM") or
Windows realm (machine$@REALM).
The name field also specifies
a fully qualified domain name. This often
represents the name of the record to be updated.
Interpretation of this field is dependent on rule type.
If no types are explicitly specified,
then a rule matches all types except RRSIG, NS, SOA, NSEC,
and NSEC3. Types may be specified by name, including
"ANY"; ANY matches all types except NSEC and NSEC3,
which can never be updated. Note that when an attempt
is made to delete all records associated with a name,
the rules are checked for each existing record type.
The ruletype field has 16
values:
name, subdomain,
zonesub, wildcard,
self, selfsub,
selfwild, ms-self,
ms-selfsub, ms-subdomain,
krb5-self, krb5-selfsub,
krb5-subdomain, tcp-self,
6to4-self, and external.
When multiple views are in use, a zone may be
referenced by more than one of them. Often, the views
contain different zones with the same name, allowing
different clients to receive different answers for the same
queries. At times, however, it is desirable for multiple
views to contain identical zones. The
in-view zone option provides an efficient
way to do this; it allows a view to reference a zone that
was defined in a previously configured view. Example:
view internal {
match-clients { 10/8; };
zone example.com {
type master;
file "example-external.db";
};
};
view external {
match-clients { any; };
zone example.com {
in-view internal;
};
};
An in-view option cannot refer to a view
that is configured later in the configuration file.
A zone statement which uses the
in-view option may not use any other
, with the exception of forward
and forwarders. (These options control
the behavior of the containing view, rather than change
the zone object itself.)
Zone-level ACLs (e.g., allow-query, allow-transfer), and
other configuration details of the zone, are all set
in the view the referenced zone is defined in. Be
careful to ensure that ACLs are wide enough
for all views referencing the zone.
An in-view zone cannot be used as a
response policy zone.
An in-view zone is not intended to reference
a forward zone.